Search Results

Search found 105 results on 5 pages for 'affinity'.

Page 3/5 | < Previous Page | 1 2 3 4 5  | Next Page >

• Reset Network Load Balancer Connection Pool

  - by bill_the_loser

  I am currently working on load testing a web application on a virtual machine cluster. I am looking for a way to flush the connection pool / NLB cache so that it is like each machine connecting to the NLB is connecting for the first time and doesn't get directed back to the node that it was on last time. This a Windows 2003 Server cluster, behind a Microsoft software based Network Load Balancer. Additional Information: To do the load testing I'm using virtual machines, one for each node on the cluster. Somehow I got two virtual machines connecting to the same node and I'm looking for an easier way to reset those connections without going in to the NLB Manager and stopping and starting each node on the NLB. Update: We went ahead and changed the affinity on all of the nodes of the cluster to none. Now it's a non-issue.

  Read the article

• On VMWare Server, how can I specify which Ethernet port should be the one to be bridged?

  - by DrDavid

  My server has 4 ethernet ports; 1 uses a APIPA address to connect to a DroboElite (that's how they're designed to work), 1 is connected to the LAN, and the other two are unused currently. The issue is that VMWare Server seems to have an affinity to the port that's connected to the Drobo, which means that it can never reach the internet. If I disable the Drobo port, everything works just fine. If I enable the drobo port, nothing works (well, the drobo works, but the virtual machine doesn't ;) ) How can I tell VMWare Server to NOT use the drobo port when I'm bridging the connection?

  Read the article

• Web Farm Framework - Missing IIS features

  - by Buginator

  I'm trying to install a web farm using Microsoft's Web Farm Framework 2.2. The server is Windows 2008 R2 with IIS 7.5. I followed a tutorial. Installed WFF from Web Platform Installer. However, I'm missing some key features in the "Server Farm" panel in IIS. This is how my setup looks like However, just like in the tutorial, I want it like this How can I enable ALL the things, like Load Balancer, Health Test, Server Affinity etc? Thanks. The tutorial I used was this: weblogs.asp.net/scottgu/archive/2010/09/08/introducing-the-microsoft-web-farm-framework.aspx

  Read the article

• Audio splitting and noise removal on Windows

  - by pts

  My mother has about 100 hours of audio in a mix of MP3 and WAV files, the digitized versions of her vinyl records. Each file contains about 5 songs with a few seconds of (noisy) pause between them. My mother needs software for Windows XP with which she can listen to the files, find the gaps manually, split the files at the gaps found, reduce noise on each song, and export the songs to individual MP3 files. My mother has very limited software user skills and affinity, and she doesn't speak English. The simpler the software, the better for her, even if noise reduction is worse than with a more sophisticated, but more complicated software. I'd prefer free software, freeware or shareware (which can do all above). Please recommend something much simpler than Audacity. The software should guide the user through the process, always showing the next few available steps, and being intuitive in the sense that there are only a few allowed actions and it's obvious what they are and how to activate them. Which software would you recommend?

  Read the article

• How to have a soft-real-time process in presense of heavily swapping IO-intensive background load?

  - by Vi

  schedtool: PID 32301: PRIO 4, POLICY R: SCHED_RR , NICE -20, AFFINITY 0xf ionice: realtime: prio 4 But the music is stumbling anyway. Background load is low prio (SCHED_IDLEPRIO, idle ionice), but uses a lot of memory (more than is physically available) and does a lot of IO and calculations. Latencytop shows about 1500ms for: Following symlink Writing buffer to disk (sync) Page fault Writing a page to disk both for the bg load and for unrelated processes. Load average is 10 and counting. Why cannot it allocate, for example, 200MHZ of one of the cores and 32M of memory and not less than once per second opportunity for IO for mplayer to make it happy while continuing calculations on the background? Or: why it cannot leave background task and swap loving each other but keeping the rest of the system as if there were no background load? How to have RT processes AND heavy bg load simultaneously (without of virtual machines)?

  Read the article

• Can't browse computer via nlb cluster name

  - by peg_leg

  I have a fileserver nlb cluster, currently set to single affinity, made up of 2 2008R2 servers. We switched the primary node today. Now our Windows XP workstations can't browse to the cluster name (i.e. \fileserver) but can browse to the cluster ip address (i.e. \192.168.1.1) and can browse the member server by name (i.e. \filesvr1). I remember having a similar issue when we had to change a registry setting to allow Windows XP boxes to see another file server that was in a failover cluster but had to be referred to by another name (\thisfileserver instead of \fileserver). Convoluted, for sure, but it helped to prevent any code changes from happening. Well all of the programmers have their code on \fileserver and we can't have them switch their links every time \filesvr1 supercedes \filesvr2 or vice versa. I can't remember that registry setting that allowed the file server to ignore that it's being called by the wrong name. HELP!

  Read the article

• Setting CPU cores off-limits to all threads not specified (preferably in Windows 7)

  - by Shinrai

  I have a really specific machine configuration in the works that would really be helped out if there were any way to do this...basically what I'm looking for is the opposite of setting CPU Affinity for a process. I want to be able to tell Windows "No applications except [x] are allowed on [these cores]." Is there any mechanism whatsoever for doing this? (Yes, I am aware of some of the potential issues this could cause and I normally would never fool with processor affinities, since the OS usually does a damned good job itself, but this is a pretty odd situation involving some software that is very CPU-bound constantly having to wait on interrupts and DPCs and things from other threads.)

  Read the article

• Restrict whole system on certain cores except a few process?

  - by icando

  Hi I am running some latency sensitive program on a Linux machine (more specifically, CentOS 6), and I don't want the threads of the process being preempted. So in my plan, the first step is to set cpu affinity of the threads so that threads are running on separate cores, so they don't preempt each other. Then the second step is to make sure other processes in the system not running on these cores. So my question is: is it possible to restrict the whole system running on certain cores, except this process? This should apply to any newly created processes in the future.

  Read the article

• Code Bubbles: Disruption comes to the IDE

  - by andrewbrust

  If you’re like me, you might see the open source Eclipse IDE as a copy or, more generously, a port of the Microsoft’s Visual Studio for the non-.NET world.  It’s not that Microsoft invented the IDE (I would credit Borland with that), but they really took the idea and ran with it for the first version of Visual Studio .NET in 2002.  The question is whether someone outside of Microsoft could take the modern IDE yet another major step forward in both principle and productivity. I think that has actually happened already, and I think the innovator in question is a second-year Computer Science PhD student at Brown, named Andrew Bragdon.  His project, which he calls Code Bubbles, is an IDE that allows for editing, debugging and exploration of code in “bubbles” which remind me a little bit of the discrete note tiles on OneNote…but they’re much more than that.  Bubbles actually allow for call stack traversal, saved debug sessions, sophisticated breakpoint and value watch behaviors and more.  And because bubbles, unlike windows, are borderless, and focus on code fragments rather than whole files, the de-cluttering effect is unbelievably liberating.  The best way to understand what Code Bubbles does is to watch the screencast video:     Code Bubbles is an IDE for Java development.  Why didn’t Microsoft come up with something like this for .NET devs?  Between the existing features in Visual Studio 2010, its WPF code editor, and the fact that OneNote’s UI bears some affinity to Code Bubbles’, it’s interesting that Microsoft still has not thought outside of its own “box” to get us something like this. Heck, that’s easy for me to say.  But it’s easy for you to say that you’d like something like this in Visual Studio sometime soon.  That’s because the ASP.NET site within UserVoice is taking votes on this very issue.  Just click this link and vote! Thanks to my fellow Microsoft Regional Director Sondre Bjellås for making me aware of Code Bubbles, and to RD Steve Smith for creating the UserVoice voting option.

  Read the article

• Documentation in Oracle Retail Analytics, Release 13.3

  - by Oracle Retail Documentation Team

  The 13.3 Release of Oracle Retail Analytics is now available on the Oracle Software Delivery Cloud and from My Oracle Support. The Oracle Retail Analytics 13.3 release introduced significant new functionality with its new Customer Analytics module. The Customer Analytics module enables you to perform retail analysis of customers and customer segments. Market basket analysis (part of the Customer Analytics module) provides insight into which products have strong affinity with one another. Customer behavior information is obtained from mining sales transaction history, and it is correlated with customer segment attributes to inform promotion strategies. The ability to understand market basket affinities allows marketers to calculate, monitor, and build promotion strategies based on critical metrics such as customer profitability. Highlighted End User Documentation Updates With the addition of Oracle Retail Customer Analytics, the documentation set addresses both modules under the single umbrella name of Oracle Retail Analytics. Note, however, that the modules, Oracle Retail Merchandising Analytics and Oracle Retail Customer Analytics, are licensed separately. To accommodate new functionality, the Retail Analytics suite of documentation has been updated in the following areas, among others: The User Guide has been updated with an overview of Customer Analytics. It also contains a list of metrics associated with Customer Analytics. The Operations Guide provides details on Market Basket Analysis as well as an updated list of APIs. The program reference list now also details the module (Merchandising Analytics or Customer Analytics) to which each program applies. The Data Model was updated to include new information related to Customer Analytics, and a new section, Market Basket Analysis Module, was added to the document with its own entity relationship diagrams and data definitions. List of Documents The following documents are included in Oracle Retail Analytics 13.3: Oracle Retail Analytics Release Notes Oracle Retail Analytics Installation Guide Oracle Retail Analytics User Guide Oracle Retail Analytics Implementation Guide Oracle Retail Analytics Operations Guide Oracle Retail Analytics Data Model

  Read the article

• Custom Session Management using HashTable

  - by kaleidoscope

  ASP.NET session state lets you associate a server-side string or object dictionary containing state data with a particular HTTP client session. A session is defined as a series of requests issued by the same client within a certain period of time, and is managed by associating a session ID with each unique client. The ID is supplied by the client on each request, either in a cookie or as a special fragment of the request URL. The session data is stored on the server side in one of the supported session state stores, which include in-process memory, SQL Server™ database, and the ASP.NET State Server service. The latter two modes enable session state to be shared among multiple Web servers on a Web farm and do not require server affinity. Implement Custom session Handler you need to follow following process : 1. Create class library which will inherit from  SessionStateStoreProviderBase abstract Class. 2. Implement all abstract Method in your base class. 3.Change Mode of session to “Custom” in web.config file and provide Provider as your Namespace with classname. <sessionState mode=”Custom” customProvider=”Namespace.classname”> <Providers> <add name=”Name” type=”Namespace.classname”> </sessionstate> For more Details Please refer following links :   http://msdn.microsoft.com/en-us/magazine/cc163730.aspx http://msdn.microsoft.com/en-us/library/system.web.sessionstate.sessionstatestoreproviderbase.aspx - Chandraprakash, S Technorati Tags: Chandraprakash,Session state Managment

  Read the article

• New SQLOS features in SQL Server 2012

  - by SQLOS Team

  Here's a quick summary of SQLOS feature enhancements going into SQL Server 2012. Most of these are already in the CTP3 pre-release, except for the Resource Governor enhancements which will be in the release candidate. We've blogged about a couple of these items before. I plan to add detail. Let me know which ones you'd like to see more on: - Memory Manager Redesign: Predictable sizing and governing SQL memory consumption: sp_configure ‘max server memory’ now limits all memory committed by SQL ServerResource Governor governs all SQL memory consumption (other than special cases like buffer pool) Improved scalability of complex queries and operations that make >8K allocations Improved CPU and NUMA locality for memory accesses Single memory manager that handles page allocations of all sizes Consistent Out-of-memory handling & management across different internal components - Optimized Memory Broker for Column Store indexes (Project Apollo) - Resource Governor Support larger scale multi-tenancy by increasing Max. number of resource pools20 -> 64 [for 64-bit] Enable predictable chargeback and isolation by adding a hard cap on CPU usage Enable vertical isolation of machine resources Resource pools can be affinitized to individual or groups of schedulers or to NUMA nodes New DMV for resource pool affinity  - CLR 4 support, adds .NET Framework 4 advantages - sp_server_dianostics Captures diagnostic data and health information about SQL Server to detect potential failures Analyze internal system state Reliable when nothing else is working   - New SQLOS DMVs (in 2008 R2SP1) SQL Server related configuration - New DMVsys.dm_server_services OS related resource configurationNew DMVssys.dm_os_volume_statssys.dm_os_windows_infosys.dm_server_registry XEvents for SQL and OS related Perfmon counters Extend sys.dm_os_sys_info See previous blog posts here and here. - Scale / Mission critical Increased scalability: Support Windows 8 max memory and logical processorsDynamic Memory support in Standard Edition - Hot-Add Memory enabled when virtualized - Various Tier1 Performance Improvements, including reduced instructions for superlatches. Originally posted at http://blogs.msdn.com/b/sqlosteam/

  Read the article

• UML Diagrams of Multi-Threaded Applications

  - by PersonalNexus

  For single-threaded applications I like to use class diagrams to get an overview of the architecture of that application. This type of diagram, however, hasn’t been very helpful when trying to understand heavily multi-threaded/concurrent applications, for instance because different instances of a class "live" on different threads (meaning accessing an instance is save only from the one thread it lives on). Consequently, associations between classes don’t necessarily mean that I can call methods on those objects, but instead I have to make that call on the target object's thread. Most literature I have dug up on the topic such as Designing Concurrent, Distributed, and Real-Time Applications with UML by Hassan Gomaa had some nice ideas, such as drawing thread boundaries into object diagrams, but overall seemed a bit too academic and wordy to be really useful. I don’t want to use these diagrams as a high-level view of the problem domain, but rather as a detailed description of my classes/objects, their interactions and the limitations due to thread-boundaries I mentioned above. I would therefore like to know: What types of diagrams have you found to be most helpful in understanding multi-threaded applications? Are there any extensions to classic UML that take into account the peculiarities of multi-threaded applications, e.g. through annotations illustrating that some objects might live in a certain thread while others have no thread-affinity; some fields of an object may be read from any thread, but written to only from one; some methods are synchronous and return a result while others are asynchronous that get requests queued up and return results for instance via a callback on a different thread.

  Read the article

• "Oracle Coherence 3.5" Book - My Humble Review

  - by [email protected]

    After reviewing the book in more detail I say again that it is a great guide for sure. Lots of important concepts that sometimes can be somewhat confusing are deeply reviewed, including all types of caching schemes and backing maps, and the cache topologies with their corresponding performances and very useful "When to use it?" sections. Some functionalities that are very desirable or used a lot are reviewed with examples and best practices of implementation, including: Data affinity Querying Pagination Indexes Aggregations Event processing, listening and triggering Data persistence Security Regarding the networking and architecture topics, Coherence*Extend is exhaustively reviewed, including C++ and .NET clients, with very good tips and examples, even including source codes. Personally, I am also glad to see that the address providers (<address-provider> tag), new feature in Coherence 3.5 which is a way to programmatically provide well-known addresses in order to connect to the cluster, is mentioned on the book, because it provides new functionalities to satisfy some special configuration requirements for example: Provide a way to switch extend nodes in cases of failure Implement custom load balancing algorithms and/or dynamic discovery of TCP/IP connection acceptors Dynamically assign TCP address and port settings when binding to a server socket Another very interesting and useful section is the "Coherent Bank Sample Application", which is a great tutorial, useful to understand how Coherence interacts with third party products establishing a clear integration with them, including the use of non-Oracle products like MS Visual Studio.  

  Read the article

• Can Dungeons & Dragons Make You More Successful? [Video]

  - by Jason Fitzpatrick

  Dungeons & Dragons gets a bit of a bad rap in popular culture, but in this video treatise from Idea Channel, they propose that Dungeons & Dragons wires players for success. There are some deeply ingrained stereotypes about Dungeons & Dragons, and those stereotypes usually begin and end with people shouting “NERD!!!” But the reality of the D&D universe is a whole lot more complex. Rather than being an escape from reality, D&D is actually a way to enhance some important real life skillz! It’s a chance to learn problem solving, visualization, interaction, organization, people management… the list could go on and on. Plus, there are some very famous non-nerds who have declared an affinity for D&D, so best stop criticizing and join in if you want to be a successful at the game of life. While we’re trying not to let our love of all things gaming cloud our judgement, we’re finding it difficult to disagree with the premise that open-ended play fosters creative and adaptive thinking. Can Dungeons & Dragons Make You A Confident & Successful Person? [via Boing Boing] HTG Explains: What is the Windows Page File and Should You Disable It? How To Get a Better Wireless Signal and Reduce Wireless Network Interference How To Troubleshoot Internet Connection Problems

  Read the article

• Dealing with Fine-Grained Cache Entries in Coherence

  - by jpurdy

  On occasion we have seen significant memory overhead when using very small cache entries. Consider the case where there is a small key (say a synthetic key stored in a long) and a small value (perhaps a number or short string). With most backing maps, each cache entry will require an instance of Map.Entry, and in the case of a LocalCache backing map (used for expiry and eviction), there is additional metadata stored (such as last access time). Given the size of this data (usually a few dozen bytes) and the granularity of Java memory allocation (often a minimum of 32 bytes per object, depending on the specific JVM implementation), it is easily possible to end up with the case where the cache entry appears to be a couple dozen bytes but ends up occupying several hundred bytes of actual heap, resulting in anywhere from a 5x to 10x increase in stated memory requirements. In most cases, this increase applies to only a few small NamedCaches, and is inconsequential -- but in some cases it might apply to one or more very large NamedCaches, in which case it may dominate memory sizing calculations. Ultimately, the requirement is to avoid the per-entry overhead, which can be done either at the application level by grouping multiple logical entries into single cache entries, or at the backing map level, again by combining multiple entries into a smaller number of larger heap objects. At the application level, it may be possible to combine objects based on parent-child or sibling relationships (basically the same requirements that would apply to using partition affinity). If there is no natural relationship, it may still be possible to combine objects, effectively using a Coherence NamedCache as a "map of maps". This forces the application to first find a collection of objects (by performing a partial hash) and then to look within that collection for the desired object. This is most naturally implemented as a collection of entry processors to avoid pulling unnecessary data back to the client (and also to encapsulate that logic within a service layer). At the backing map level, the NIO storage option keeps keys on heap, and so has limited benefit for this situation. The Elastic Data features of Coherence naturally combine entries into larger heap objects, with the caveat that only data -- and not indexes -- can be stored in Elastic Data.

  Read the article

• How-To: Run CMSDK against a RAC cluster

  - by frank.closheim

  Using CMSDK in a production environment often requires a robust, reliable and failover enabled repository. When using Oracle Real Application Cluster (RAC) with your CMSDK repository you need to have a specific configuration in place to support such a setup. This post will explain the configuration steps required when running CMSDK 9.0.4.6 with Oracle WebLogic Server (WLS).In the previous CMSDK 9.0.4.2 version a RAC enabled connect string looked like this: (DESCRIPTION = (ADDRESS = (PROTOCOL = TCP)(HOST = rac1)(PORT = 1521))(ADDRESS = (PROTOCOL = TCP)(HOST = rac2)(PORT = 1521))(LOAD_BALANCE = NO)(FAILOVER = ON)(CONNECT_DATA =(SERVICE_NAME = rac)(failover_mode = (type=select)(method=basic)))CMSDK 9.0.4.6 makes use of data sources to connect to the underlying database. These data sources are configured inside your Application Server, such as Oracle WebLogic Server.In Oracle WebLogic Server 10.3.4, a single data source implementation has been introduced to support an RAC cluster. It responds to Fast Application Notification (FAN) events to provide Fast Connection Failover (FCF), Runtime Connection Load-Balancing (RCLB), and RAC instance graceful shutdown. XA affinity is supported at the global transaction Id level. The new feature is called WebLogic Active GridLink for RAC; which is implemented as the GridLink data source within WebLogic Server.This GridLink data source also works with Oracle Single Client Access Name (SCAN). SCAN is a feature used in RAC environments that provides a single name for clients to access any Oracle Database running in a cluster. You can think of SCAN as a cluster alias for databases in the cluster. The benefit is that the client’s connect information does not need to change if you add or remove nodes or databases in the cluster.The CMSDK 9.0.4.6 documentation describes how to create a regular JDBC data source named jdbc/OracleDS. Please refer to the following document which describes in detail how to create a GridLink data source in WLS.

  Read the article

• Context switches much slower in new linux kernels

  - by Michael Goldshteyn

  We are looking to upgrade the OS on our servers from Ubuntu 10.04 LTS to Ubuntu 12.04 LTS. Unfortunately, it seems that the latency to run a thread that has become runnable has significantly increased from the 2.6 kernel to the 3.2 kernel. In fact the latency numbers we are getting are hard to believe. Let me be more specific about the test. We have a program that has two threads. The first thread gets the current time (in ticks using RDTSC) and then signals a condition variable once a second. The second thread waits on the condition variable and wakes up when it is signaled. It then gets the current time (in ticks using RDTSC). The difference between the time in the second thread and the time in the first thread is computed and displayed on the console. After this the second thread waits on the condition variable once more. So, we get a thread to thread signaling latency measurement once a second as a result. In linux 2.6.32, this latency is somewhere on the order of 2.8-3.5 us, which is reasonable. In linux 3.2.0, this latency is somewhere on the order of 40-100 us. I have excluded any differences in hardware between the two host hosts. They run on identical hardware (dual socket X5687 {Westmere-EP} processors running at 3.6 GHz with hyperthreading, speedstep and all C states turned off). We are changing the affinity to run both threads on physical cores of the same socket (i.e., the first thread is run on Core 0 and the second thread is run on Core 1), so there is no bouncing of threads on cores or bouncing/communication between sockets. The only difference between the two hosts is that one is running Ubuntu 10.04 LTS with kernel 2.6.32-28 (the fast context switch box) and the other is running the latest Ubuntu 12.04 LTS with kernel 3.2.0-23 (the slow context switch box). Have there been any changes in the kernel that could account for this ridiculous slow down in how long it takes for a thread to be scheduled to run?

  Read the article

• Usability - How to edit favorites?

  - by Florian

  Hi, I'd like to get some opinions about about usability in the following case: Target group people from 30-50, low to middle internet affinity. App: I have a website with login. Visitors can save interesseting pages in their fav-box for fast access. Here the actual question: How to edit this favorites? Is it better to give the visitors direct access to drag/dropn and delete their favs or is it better to have an edit button so they have to activate the edit mode before? The fav-link would look like this | link text to click | icon-drag | icon-delete | thx for input TC

  Read the article

• Throttle CPU Usage consumed by Process

  - by Brett Powell

  We run a game-server company where we basically have large amounts of customers sharing a single machine, and are just on their own instance of a Java Process (Minecraft) managed by our Web Control Panels. In the last few game updates released, we have noticed that many of the third-party plugins our customer's use have become poorly written and we are frequently seeing huge CPU increases from certain servers until we manually kill the process. Our Game Panel automatically restarts processes, so killing them is not really an issue. Our problem is that once once of these servers starts consuming 50%+ CPU Usage, it takes atleast 5 minutes to RDP into the machine, locate who it belongs to, shut it down and notify them. Are there any current solutions for Server 2008 which allow for the throttling of CPU usage or worst case, just auto kill a process stuck using that much? As Minecraft is essentially a single-threaded application, we have investigated using Affinity, although with the variations in our Packages and fluctuations in usage, this doesn't work well for us. Some option to throttle the maximum usage a process can use would be perfect, or at least the option to kill a process using that much. Thanks!

  Read the article

• Question for Vim search and peck typists

  - by mike

  I'm trying to write a Vim tutorial and I'd like to start by dismissing a few misconceptions, as well as giving some recommendations. I don't know if I should dismiss touch-typing as a misconception, or include it as a recommended prerequisite. At the time I learned the editor, I had already been touch typing for a couple of years, so I have absolutely no idea what would be the experience of a two-fingered typist in Vim. Are you a vim two-fingered typist? what has your experience been like? EDIT: I'm not sure if my question was clear enough. Maybe it's my fault, I don't know. I get mixed replies and other questions (why do you write this? what does one have to do with the other?), instead of empirical info (I don't touch type and it's been (fine|hell)). Some programmers touch-type others search and peck. In the middle, there's Vim which requires a certain affinity with keys to do various operations. I am a touch typist and I have no clue what my experience would have been like with the editor if I wasn't. I can't honestly picture myself pecking some of these combos. But like I said, I don't know what it is like. Before telling someone to start using Vim, I'd like to know if I should dismiss touch-typing as a misconceived requirement. So, I'll rephrase the question, have you felt that not being a touch-typist has impeded on your experience with Vim?

  Read the article

• Very uneven CPU utilization with SQL Server 2012 on 2 processor computer with 16 cores / processor

  - by cooplarsh

  After installing SQL Server Enterprise 2012 with the Server + Cal license model, on a computer with 2 processors each with 16 cores (and no hyperthreading involved) and putting the server under extremely heavy load the 16 cores on the first processor were very underutilized, the first 4 cores on the 2nd CPU were heavily utilized, and the last 12 cores were not used at all (because of the 20 core limit for this sql server version). Total CPU utilization was displaying as around 25%. Unfortunately, the server suffered from extremely poor performance even though if the tasks were evenly distributed across the 20 cores it wouldn't have been anywhere near as bad. The Windows Server was running on a VMWare virtual image under ESX Server, but all of the CPU was allocated to the windows server. We tried changing affinity settings (e.g., allocating most cores to CPU and the others to I/O), but that didn't help solve the performance problems. Upgrading the product edition to SQL Server Enterprise Core 2012 not only allowed the SQL Server to utilize the 12 previously unused cores on the 2nd processor, but it also resulted in a much more even distribution of tasks across all of the processors. To get through the backlog of requests cpU utilization jumped to around 90%, and then came down to around 33% once it was caught up, but performance improved dramatically since we failed over to the newly updated version And the performance issues went away. I was wondering if anyone knows what might cause SQL Server to unevenly distribute the load, relying almost exclusively on the first 4 cores of the 2nd processor that had 12 cores idle, and allocate only a few tasks to each of the 16 cores on the first processor. Also, is there any way we could have more evenly distributed the load across the 20 cores that were being used without the product edition upgrade? The flip side of that question is what did the product upgrade do that caused SQL Server to start evenly distributing the load across all of the cores that it recognized? Thanks to any insight to answer these questions and/or links that might help me better understand how to make sense of what was happenings.

  Read the article

• SQL Server performance on VSphere 4.0

  - by Charles

  We are having a performance issue that we cannot explain with our VMWare environment and I am hoping someone here may be able to help. We have a web application that uses a databases backend. We have an SQL 2005 Cluster setup on Windows 2003 R2 between a physical node and a virtual node. Both physical servers are identical 2950's with 2x Xeaon x5460 Quad Core CPUs and 64GB of memory, 16GB allocated to the OS. We are utilizing an iSCSI San for all cluster disks. The problem is this, when utilizing the application under a repeated stress testing that adds CPUs to the cluster nodes, the Physical node scales from 1 pCPU to 8 pCPUs, meaning we see continued performance increases. When testing the node running Vsphere, we have the expected 12% performance hit for being virtual but we still scale from 1 vCPU to 4 vCPUs like the physical but beyond this performance drops off, by the time we get to 8 vCPUs we are seeing performance numbers worse than at 4 vCPUs. Again, both nodes are configured identically in terms of hardware, Guest OS, SQL Configurations etc and there is no traffic other than the testing on the system. There are no other VMs on the virtual server so there should be no competition for resources. We have contacted VMWare for help but they have not really been any suggesting things like setting SQL Processor Affinity which, while being helpful would have the same net effect on each box and should not change our results in the least. We have looked at all of VMWare's SQL Tuning guides with regards to VSphere with no benefit, please help!

  Read the article

• Why should you choose Oracle WebLogic 12c instead of JBoss EAP 6?

  - by Ricardo Ferreira

  In this post, I will cover some technical differences between Oracle WebLogic 12c and JBoss EAP 6, which was released a couple days ago from Red Hat. This article claims to help you in the evaluation of key points that you should consider when choosing for an Java EE application server. In the following sections, I will present to you some important aspects that most customers ask us when they are seriously evaluating for an middleware infrastructure, specially if you are considering JBoss for some reason. I would suggest that you keep the following question in mind while you are reading the points: "Why should I choose JBoss instead of WebLogic?" 1) Multi Datacenter Deployment and Clustering - D/R ("Disaster & Recovery") architecture support is embedded on the WebLogic Server 12c product. JBoss EAP 6 on the other hand has no direct D/R support included, Red Hat relies on third-part tools with higher prices. When you consider a middleware solution to host your business critical application, you should worry with every architectural aspect that are related with the solution. Fail-over support is one little aspect of a truly reliable solution. If you do not worry about D/R, your solution will not be reliable. Having said that, with Red Hat and JBoss EAP 6, you have this extra cost that will increase considerably the total cost of ownership of the solution. As we commonly hear from analysts, open-source are not so cheaper when you start seeing the big picture. - WebLogic Server 12c supports advanced LAN clustering, detection of death servers and have a common alert framework. JBoss EAP 6 on the other hand has limited LAN clustering support with no server death detection. They do not generate any alerts when servers goes down (only if you buy JBoss ON which is a separated technology, but until now does not support JBoss EAP 6) and manual intervention are required when servers goes down. In most cases, admin people must rely on "kill -9", "tail -f someFile.log" and "ps ax | grep java" commands to manage failures and clustering anomalies. - WebLogic Server 12c supports the concept of Node Manager, which is a separated process that runs on the physical | virtual servers that allows extend the administration of the cluster to WebLogic managed servers that are often distributed across multiple machines and geographic locations. JBoss EAP 6 on the other hand has no equivalent technology. Whole server instances must be managed individually. - WebLogic Server 12c Node Manager supports Coherence to boost performance when managing servers. JBoss EAP 6 on the other hand has no similar technology. There is no way to coordinate JBoss and infiniband instances provided by JBoss using high throughput and low latency protocols like InfiniBand. The Node Manager feature also allows another very important feature that JBoss EAP lacks: secure the administration. When using WebLogic Node Manager, all the administration tasks are sent to the managed servers in a secure tunel protected by a certificate, which means that the transport layer that separates the WebLogic administration console from the managed servers are secured by SSL. - WebLogic Server 12c are now integrated with OTD ("Oracle Traffic Director") which is a web server technology derived from the former Sun iPlanet Web Server. This software complements the web server support offered by OHS ("Oracle HTTP Server"). Using OTD, WebLogic instances are load-balanced by a high powerful software that knows how to handle SDP ("Socket Direct Protocol") over InfiniBand, which boost performance when used with engineered systems technologies like Oracle Exalogic Elastic Cloud. JBoss EAP 6 on the other hand only offers support to Apache Web Server with custom modules created to deal with JBoss clusters, but only across standard TCP/IP networks.  2) Application and Runtime Diagnostics - WebLogic Server 12c have diagnostics capabilities embedded on the server called WLDF ("WebLogic Diagnostic Framework") so there is no need to rely on third-part tools. JBoss EAP 6 on the other hand has no diagnostics capabilities. Their only diagnostics tool is the log generated by the application server. Admin people are encouraged to analyse thousands of log lines to find out what is going on. - WebLogic Server 12c complement WLDF with JRockit MC ("Mission Control"), which provides to administrators and developers a complete insight about the JVM performance, behavior and possible bottlenecks. WebLogic Server 12c also have an classloader analysis tool embedded, and even a log analyzer tool that enables administrators and developers to view logs of multiple servers at the same time. JBoss EAP 6 on the other hand relies on third-part tools to do something similar. Again, only log searching are offered to find out whats going on. - WebLogic Server 12c offers end-to-end traceability and monitoring available through Oracle EM ("Enterprise Manager"), including monitoring of business transactions that flows through web servers, ESBs, application servers and database servers, all of this with high deep JVM analysis and diagnostics. JBoss EAP 6 on the other hand, even using JBoss ON ("Operations Network"), which is a separated technology, does not support those features. Red Hat relies on third-part tools to provide direct Oracle database traceability across JVMs. One of those tools are Oracle EM for non-Oracle middleware that manage JBoss, Tomcat, Websphere and IIS transparently. - WebLogic Server 12c with their JRockit support offers a tool called JRockit Flight Recorder, which can give developers a complete visibility of a certain period of application production monitoring with zero extra overhead. This automatic recording allows you to deep analyse threads latency, memory leaks, thread contention, resource utilization, stack overflow damages and GC ("Garbage Collection") cycles, to observe in real time stop-the-world phenomenons, generational, reference count and parallel collects and mutator threads analysis. JBoss EAP 6 don't even dream to support something similar, even because they don't have their own JVM. 3) Application Server Administration - WebLogic Server 12c offers a complete administration console complemented with scripting and macro-like recording capabilities. A single WebLogic console can managed up to hundreds of WebLogic servers belonging to the same domain. JBoss EAP 6 on the other hand has a limited console and provides a XML centric administration. JBoss, after ten years, started the development of a rudimentary centralized administration that still leave a lot of administration tasks aside, so admin people and developers must touch scripts and XML configuration files for most advanced and even simple administration tasks. This lead applications to error prone and risky deployments. Even using JBoss ON, JBoss EAP are not able to offer decent administration features for admin people which must be high skilled in JBoss internal architecture and its managing capabilities. - Oracle EM is available to manage multiple domains, databases, application servers, operating systems and virtualization, with a complete end-to-end visibility. JBoss ON does not provide management capabilities across the complete architecture, only basic monitoring. Even deployment must be done aside JBoss ON which does no integrate well with others softwares than JBoss. Until now, JBoss ON does not supports JBoss EAP 6, so even their minimal support for JBoss are not available for JBoss EAP 6 leaving customers uncovered and subject to high skilled JBoss admin people. - WebLogic Server 12c has the same administration model whatever is the topology selected by the customer. JBoss EAP 6 on the other hand differentiates between two operational models: standalone-mode and domain-mode, that are not consistent with each other. Depending on the mode used, the administration skill is different. - WebLogic Server 12c has no point-of-failures processes, and it does not need to define any specialized server. Domain model in WebLogic is available for years (at least ten years or more) and is production proven. JBoss EAP 6 on the other hand needs special processes to garantee JBoss integrity, the PC ("Process-Controller") and the HC ("Host-Controller"). Different from WebLogic, the domain model in JBoss is quite new (one year at tops) of maturity, and need to mature considerably until start doing things like WebLogic domain model does. - WebLogic Server 12c supports parallel deployment model which enables some artifacts being deployed at the same time. JBoss EAP 6 on the other hand does not have any similar feature. Every deployment are done atomically in the containers. This means that if you have a huge EAR (an EAR of 120 MB of size for instance) and deploy onto JBoss EAP 6, this EAR will take some minutes in order to starting accept thread requests. The same EAR deployed onto WebLogic Server 12c will reduce the deployment time at least in 2X compared to JBoss. 4) Support and Upgrades - WebLogic Server 12c has patch management available. JBoss EAP 6 on the other hand has no patch management available, each JBoss EAP instance should be patched manually. To achieve such feature, you need to buy a separated technology called JBoss ON ("Operations Network") that manage this type of stuff. But until now, JBoss ON does not support JBoss EAP 6 so, in practice, JBoss EAP 6 does not have this feature. - WebLogic Server 12c supports previuous WebLogic domains without any reconfiguration since its kernel is robust and mature since its creation in 1995. JBoss EAP 6 on the other hand has a proven lack of supportability between JBoss AS 4, 5, 6 and 7. Different kernels and messaging engines were implemented in JBoss stack in the last five years reveling their incapacity to create a well architected and proven middleware technology. - WebLogic Server 12c has patch prescription based on customer configuration. JBoss EAP 6 on the other hand has no such capability. People need to create ticket supports and have their installations revised by Red Hat support guys to gain some patch prescription from them. - Oracle WebLogic Server independent of the version has 8 years of support of new patches and has lifetime release of existing patches beyond that. JBoss EAP 6 on the other hand provides patches for a specific application server version up to 5 years after the release date. JBoss EAP 4 and previous versions had only 4 years. A good question that Red Hat will argue to answer is: "what happens when you find issues after year 5"?  5) RAC ("Real Application Clusters") Support - WebLogic Server 12c ships with a specific JDBC driver to leverage Oracle RAC clustering capabilities (Fast-Application-Notification, Transaction Affinity, Fast-Connection-Failover, etc). Oracle JDBC thin driver are also available. JBoss EAP 6 on the other hand ships only the standard Oracle JDBC thin driver. Load balancing with Oracle RAC are not supported. Manual intervention in case of planned or unplanned RAC downtime are necessary. In JBoss EAP 6, situation does not reestablish automatically after downtime. - WebLogic Server 12c has a feature called Active GridLink for Oracle RAC which provides up to 3X performance on OLTP applications. This seamless integration between WebLogic and Oracle database enable more value added to critical business applications leveraging their investments in Oracle database technology and Oracle middleware. JBoss EAP 6 on the other hand has no performance gains at all, even when admin people implement some kind of connection-pooling tuning. - WebLogic Server 12c also supports transaction and web session affinity to the Oracle RAC, which provides aditional gains of performance. This is particularly interesting if you are creating a reliable solution that are distributed not only in an LAN cluster, but into a different data center. JBoss EAP 6 on the other hand has no such support. 6) Standards and Technology Support - WebLogic Server 12c is fully Java EE 6 compatible and production ready since december of 2011. JBoss EAP 6 on the other hand became fully compatible with Java EE 6 only in the community version after three months, and production ready only in a few days considering that this article was written in June of 2012. Red Hat says that they are the masters of innovation and technology proliferation, but compared with Oracle and even other proprietary vendors like IBM, they historically speaking are lazy to deliver the most newest technologies and standards adherence. - Oracle is the steward of Java, driving innovation into the platform from commercial and open-source vendors. Red Hat on the other hand does not have its own JVM and relies on third-part JVMs to complete their application server offer. 95% of Red Hat customers are using Oracle HotSpot as JVM, which means that without Oracle involvement, their support are limited exclusively to the application server layer and we all know that most problems are happens in the JVM layer. - WebLogic Server 12c supports natively JDK 7, which empower developers to explore the maximum of the Java platform productivity when writing code. This feature differentiate WebLogic from others application servers (except GlassFish that are also managed by Oracle) because the usage of JDK 7 introduce such remarkable productivity features like the "try-with-resources" enhancement, catching multiple exceptions with one try block, Strings in the switch statements, JVM improvements in terms of JDBC, I/O, networking, security, concurrency and of course, the most important feature of Java 7: native support for multiple non-Java languages. More features regarding JDK 7 can be found here. JBoss EAP 6 on the other hand does not support JDK 7 officially, they comment in their community version that "Java SE 7 can be used with JBoss 7" which does not gives you any guarantees of enterprise support for JDK 7. - Oracle WebLogic Server 12c supports integration with Spring framework allowing Spring applications to use WebLogic special transaction manager, exposing bean interfaces to WebLogic MBeans to take advantage of all WebLogic monitoring and administration advantages. JBoss EAP 6 on the other hand has no special integration with Spring. In fact, Red Hat offers a suspicious package called "JBoss Web Platform" that in theory supports Spring, but in practice this package does not offers any special integration. It is just a facility for Red Hat customers to have support from both JBoss and Spring technology using the same customer support. 7) Lightweight Development - Oracle WebLogic Server 12c and Oracle GlassFish are completely integrated and can share applications without any modifications. Starting with the 12c version, WebLogic now understands natively GlassFish deployment descriptors and specific configurations in order to offer you a truly and reliable migration path from a community Java EE application server to a enterprise middleware product like WebLogic. JBoss EAP 6 on the other hand has no support to natively reuse an existing (or still in development) application from JBoss AS community server. Users of JBoss suffer of critical issues during deployment time that includes: changing the libraries and dependencies of the application, patching the DTD or XSD deployment descriptors, refactoring of the application layers due classloading issues and anomalies, rebuilding of persistence, business and web layers due issues with "usage of the certified version of an certain dependency" or "frameworks that Red Hat potentially does not recommend" etc. If you have the culture or enterprise IT directive of developing Java EE applications using community middleware to in a certain future, transition to enterprise (supported by a vendor) middleware, Oracle WebLogic plus Oracle GlassFish offers you a more sustainable solution. - WebLogic Server 12c has a very light ZIP distribution (less than 165 MB). JBoss EAP 6 ZIP size is around 130 MB, together with JBoss ON you have more 100 MB resulting in a higher download footprint. This is particularly interesting if you plan to use automated setup of application server instances (for example, to rapidly setup a development or staging environment) using Maven or Hudson. - WebLogic Server 12c has a complete integration with Maven allowing developers to setup WebLogic domains with few commands. Tasks like downloading WebLogic, installation, domain creation, data sources deployment are completely integrated. JBoss EAP 6 on the other hand has a limited offer integration with those tools.  - WebLogic Server 12c has a startup mode called WLX that turns-off EJB, JMS and JCA containers leaving enabled only the web container with Java EE 6 web profile. JBoss EAP 6 on the other hand has no such feature, you need to disable manually the containers that you do not want to use. - WebLogic Server 12c supports fastswap, which enables you to change classes without redeployment. This is particularly interesting if you are developing patches for the application that is already deployed and you do not want to redeploy the entire application. This is the same behavior that most application servers offers to JSP pages, but with WebLogic Server 12c, you have the same feature for Java classes in general. JBoss EAP 6 on the other hand has no such support. Even JBoss EAP 5 does not support this until now. 8) JMS and Messaging - WebLogic Server 12c has a proven and high scalable JMS implementation since its initial release in 1995. JBoss EAP 6 on the other hand has a still immature technology called HornetQ, which was introduced in JBoss EAP 5 replacing everything that was implemented in the previous versions. Red Hat loves to introduce new technologies across JBoss versions, playing around with customers and their investments. And when they are asked about why they have changed the implementation and caused such a mess, their answer is always: "the previous implementation was inadequate and not aligned with the community strategy so we are creating a new a improved one". This Red Hat practice leads to uncomfortable investments that in a near future (sometimes less than a year) will be affected in someway. - WebLogic Server 12c has troubleshooting and monitoring features included on the WebLogic console and WLDF. JBoss EAP 6 on the other hand has no direct monitoring on the console, activity is reflected only on the logs, no debug logs available in case of JMS issues. - WebLogic Server 12c has extremely good performance and scalability. JBoss EAP 6 on the other hand has a JMS storage mechanism relying on Oracle database or MySQL. This means that if an issue in production happens and Red Hat affirms that an performance issue is happening due to database problems, they will not support you on the performance issue. They will orient you to call Oracle instead. - WebLogic Server 12c supports messaging enterprise features like SAF ("Store and Forward"), Distributed Queues/Topics and Foreign JMS providers support that leverage JMS implementations without compromise developer code making things completely transparent. JBoss EAP 6 on the other hand do not even dream to support such features. 9) Caching and Grid - Coherence, which is the leading and most mature data grid technology from Oracle, is available since early 2000 and was integrated with WebLogic in 2009. Coherence and WebLogic clusters can be both managed from WebLogic administrative console. Even Node Manager supports Coherence. JBoss on the other hand discontinued JBoss Cache, which was their caching implementation just like they did with the messaging implementation (JBossMQ) which was a issue for long term customers. JBoss EAP 6 ships InfiniSpan version 1.0 which is immature and lack a proven record of successful cases and reliability. - WebLogic Server 12c has a feature called ActiveCache which uses Coherence to, without any code changes, replicate HTTP sessions from both WebLogic and other application servers like JBoss, Tomcat, Websphere, GlassFish and even Microsoft IIS. JBoss EAP 6 on the other hand does have such support and even when they do in the future, they probably will support only their own application server. - Coherence can be used to manage both L1 and L2 cache levels, providing support to Oracle TopLink and others JPA compliant implementations, even Hibernate. JBoss EAP 6 and Infinispan on the other hand supports only Hibernate. And most important of all: Infinispan does not have any successful case of L1 or L2 caching level support using Hibernate, which lead us to reflect about its viability. 10) Performance - WebLogic Server 12c is certified with Oracle Exalogic Elastic Cloud and can run unchanged applications at this engineered system. This approach can benefit customers from Exalogic optimization's of both kernel and JVM layers to boost performance in terms of 10X for web, OLTP, JMS and grid applications. JBoss EAP 6 on the other hand has no investment on engineered systems: customers do not have the choice to deploy on a Java ultra fast system if their project becomes relevant and performance issues are detected. - WebLogic Server 12c maintains a performance gain across each new release: starting on WebLogic 5.1, the overall performance gain has been close to 4X, which close to a 20% gain release by release. JBoss on the other hand does not provide SPECJAppServer or SPECJEnterprise performance benchmarks. Their so called "performance gains" remains hidden in their customer environments, which lead us to think if it is true or not since we will never get access to those environments. - WebLogic Server 12c has industry performance benchmarks with submissions across platforms and configurations leading SPECJ. Oracle WebLogic leads SPECJAppServer performance in multiple categories, fitting all customer topologies like: dual-node, single-node, multi-node and multi-node with RAC. JBoss... again, does not provide any SPECJAppServer performance benchmarks. - WebLogic Server 12c has a feature called work manager which allows your application to embrace new performance levels based on critical resource utilization of the CPUs usage. Work managers prioritizes work and allocates threads based on an execution model that takes into account administrator-defined parameters and actual run-time performance and throughput. JBoss EAP 6 on the other hand has no compared feature and probably they never will. Not supporting such feature like work managers, JBoss EAP 6 forces admin people and specially developers to uncover performance gains in a intrusive way, rewriting the code and doing performance refactorings. 11) Professional Services Support - WebLogic Server 12c and any other technology sold by Oracle give customers the possibility of hire OCS ("Oracle Consulting Services") to manage critical scenarios, deployment assistance of new applications, high skilled consultancy of architecture, best practices and people allocation together with customer teams. All OCS services are available without any restrictions, having the customer bought software from Oracle or just starting their implementation before any acquisition. JBoss EAP 6 or Red Hat to be more specifically, only offers professional services if you buy subscriptions from them. If you are developing a new critical application for your business and need the help of Red Hat for a serious issue or architecture decision, they will probably say: "OK... I can help you but after you buy subscriptions from me". Red Hat also does not allows their professional services consultants to manage environments that uses community based software. They will probably force you to first buy a subscription, download their "enterprise" version and them, optionally hire their consultants. - Oracle provides you our university to educate your team into our technologies, including of course specialized trainings of WebLogic application server. At any time and location, you can hire Oracle to train your team so you get trustful knowledge according to your specific needs. Certifications for the products are also available if your technical people desire to differentiate themselves as professionals. Red Hat on the other hand have a limited pool of resources to train your team in their technologies. Basically they are selling training and certification for RHEL ("Red Hat Enterprise Linux") but if you demand more specialized training in JBoss middleware, they will probably connect you to some "certified" partner localized training since they are apparently discontinuing their education center, at least here in Brazil. They were not able to reproduce their success with RHEL education to their middleware division since they need first sell the subscriptions to after gives you specialized training. And again, they only offer you specialized training based on their enterprise version (EAP in the case of JBoss) which means that the courses will be a quite outdated. There are reports of developers that took official training's from Red Hat at this year (2012) and in a certain JBoss advanced course, Red Hat supposedly covered JBossMQ as the messaging subsystem, and even the printed material provided was based on JBossMQ since the training was created for JBoss EAP 4.3. 12) Encouraging Transparency without Ulterior Motives - WebLogic Server 12c like any other software from Oracle can be downloaded any time from anywhere, you should only possess an OTN ("Oracle Technology Network") credential and you can download any enterprise software how many times you want. And is not some kind of "trial" version. It is the official binaries that will be running for ever in your data center. Oracle does not encourages the usage of "specific versions" of our software. The binaries you buy from Oracle are the same binaries anyone in the world could download and use for testing and personal education. JBoss EAP 6 on the other hand are not available for download unless you buy a subscription and get access to the Red Hat enterprise repositories. If you need to test, learn or just start creating your application using Red Hat's middleware software, you should download it from the community website. You are not allowed to download the enterprise version that, according to Red Hat are more secure, reliable and robust. But no one of us want to start the development of a software with an unsecured, unreliable and not scalable middleware right? So what you do? You are "invited" by Red Hat to buy subscriptions from them to get access to the "cool" version of the software. - WebLogic Server 12c prices are publicly available in the Oracle website. If you want to know right now how much WebLogic will cost to your organization, just click here and get access to our price list. In the case of WebLogic, check out the "US Oracle Technology Commercial Price List". Oracle also encourages you to get in touch with a sales representative to discuss discounts that would make possible the investment into our technology. But you are not required to do this, only if you are interested in buying our technology or maybe you want to discuss some discount scenarios. JBoss EAP 6 on the other hand does not have its cost publicly available in Red Hat's website or in any other media, at least is not so easy to get such information. The only link you will possibly find in their website is a "Contact a Sales Representative" link. This is not a very good relationship between an customer and an vendor. This is not an example of transparency, mainly when the software are sold as open. In this situations, customers expects to see the software prices publicly available, so they can have the chance to decide, based on the existing features of the software, if the cost is fair or not. Conclusion Oracle WebLogic is the most mature, secure, reliable and scalable Java EE application server of the market, and have a proven record of success around the globe to prove it's majority. Don't lose the chance to discover today how WebLogic could fit your needs and sustain your global IT middleware strategy, no matter if your strategy are completely based on the Cloud or not.

  Read the article

• unexplainable packet drops with 5 ethernet NICs and low traffic on Ubuntu

  - by jon

  I'm stuck on problem where my machine started to drops packets with no sign of ANY system load or high interrupt usage after an upgrade to Ubuntu 12.04. My server is a network monitoring sensor, running Ubuntu LTS 12.04, it passively collects packets from 5 interfaces doing network intrusion type stuff. Before the upgrade I managed to collect 200+GB of packets a day while writing them to disk with around 0% packet loss depending on the day with the help of CPU affinity and NIC IRQ to CPU bindings. Now I lose a great deal of packets with none of my applications running and at very low PPS rate which a modern workstation NIC would have no trouble with. Specs: x64 Xeon 4 cores 3.2 Ghz 16 GB RAM NICs: 5 Intel Pro NICs using the e1000 driver (NAPI). [1] eth0 and eth1 are integrated NICs (in the motherboard) There are 2 other PCI-X network cards, each with 2 Ethernet ports. 3 of the interfaces are running at Gigabit Ethernet, the others are not because they're attached to hubs. Specs: [2] http://support.dell.com/support/edocs/systems/pe2850/en/ug/t1390aa.htm uptime 17:36:00 up 1:43, 2 users, load average: 0.00, 0.01, 0.05 # uname -a Linux nms 3.2.0-29-generic #46-Ubuntu SMP Fri Jul 27 17:03:23 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux I also have the CPU governor set to performance mode and irqbalance off. The problem still occurs with them on. # lspci -t -vv -[0000:00]-+-00.0 Intel Corporation E7520 Memory Controller Hub +-02.0-[01-03]--+-00.0-[02]----0e.0 Dell PowerEdge Expandable RAID controller 4 | \-00.2-[03]-- +-04.0-[04]-- +-05.0-[05-07]--+-00.0-[06]----07.0 Intel Corporation 82541GI Gigabit Ethernet Controller | \-00.2-[07]----08.0 Intel Corporation 82541GI Gigabit Ethernet Controller +-06.0-[08-0a]--+-00.0-[09]--+-04.0 Intel Corporation 82546EB Gigabit Ethernet Controller (Copper) | | \-04.1 Intel Corporation 82546EB Gigabit Ethernet Controller (Copper) | \-00.2-[0a]--+-02.0 Digium, Inc. Wildcard TE210P/TE212P dual-span T1/E1/J1 card 3.3V | +-03.0 Intel Corporation 82546EB Gigabit Ethernet Controller (Copper) | \-03.1 Intel Corporation 82546EB Gigabit Ethernet Controller (Copper) +-1d.0 Intel Corporation 82801EB/ER (ICH5/ICH5R) USB UHCI Controller #1 +-1d.1 Intel Corporation 82801EB/ER (ICH5/ICH5R) USB UHCI Controller #2 +-1d.2 Intel Corporation 82801EB/ER (ICH5/ICH5R) USB UHCI Controller #3 +-1d.7 Intel Corporation 82801EB/ER (ICH5/ICH5R) USB2 EHCI Controller +-1e.0-[0b]----0d.0 Advanced Micro Devices [AMD] nee ATI RV100 QY [Radeon 7000/VE] +-1f.0 Intel Corporation 82801EB/ER (ICH5/ICH5R) LPC Interface Bridge \-1f.1 Intel Corporation 82801EB/ER (ICH5/ICH5R) IDE Controller I believe the NIC nor the NIC drivers are dropping the packets because ethtool reports 0 under rx_missed_errors and rx_no_buffer_count for each interface. On the old system, if it couldn't keep up this is where the drops would be. I drop packets on multiple interfaces just about every second, usually in small increments of 2-4. I tried all these sysctl values, I'm currently using the uncommented ones. # cat /etc/sysctl.conf # high net.core.netdev_max_backlog = 3000000 net.core.rmem_max = 16000000 net.core.rmem_default = 8000000 # defaults #net.core.netdev_max_backlog = 1000 #net.core.rmem_max = 131071 #net.core.rmem_default = 163480 # moderate #net.core.netdev_max_backlog = 10000 #net.core.rmem_max = 33554432 #net.core.rmem_default = 33554432 Here's an example of an interface stats report with ethtool. They are all the same, nothing is out of the ordinary ( I think ), so I'm only going to show one: ethtool -S eth2 NIC statistics: rx_packets: 7498 tx_packets: 0 rx_bytes: 2722585 tx_bytes: 0 rx_broadcast: 327 tx_broadcast: 0 rx_multicast: 1504 tx_multicast: 0 rx_errors: 0 tx_errors: 0 tx_dropped: 0 multicast: 1504 collisions: 0 rx_length_errors: 0 rx_over_errors: 0 rx_crc_errors: 0 rx_frame_errors: 0 rx_no_buffer_count: 0 rx_missed_errors: 0 tx_aborted_errors: 0 tx_carrier_errors: 0 tx_fifo_errors: 0 tx_heartbeat_errors: 0 tx_window_errors: 0 tx_abort_late_coll: 0 tx_deferred_ok: 0 tx_single_coll_ok: 0 tx_multi_coll_ok: 0 tx_timeout_count: 0 tx_restart_queue: 0 rx_long_length_errors: 0 rx_short_length_errors: 0 rx_align_errors: 0 tx_tcp_seg_good: 0 tx_tcp_seg_failed: 0 rx_flow_control_xon: 0 rx_flow_control_xoff: 0 tx_flow_control_xon: 0 tx_flow_control_xoff: 0 rx_long_byte_count: 2722585 rx_csum_offload_good: 0 rx_csum_offload_errors: 0 alloc_rx_buff_failed: 0 tx_smbus: 0 rx_smbus: 0 dropped_smbus: 01 # ifconfig eth0 Link encap:Ethernet HWaddr 00:11:43:e0:e2:8c UP BROADCAST RUNNING NOARP PROMISC ALLMULTI MULTICAST MTU:1500 Metric:1 RX packets:373348 errors:16 dropped:95 overruns:0 frame:16 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:356830572 (356.8 MB) TX bytes:0 (0.0 B) eth1 Link encap:Ethernet HWaddr 00:11:43:e0:e2:8d UP BROADCAST RUNNING NOARP PROMISC ALLMULTI MULTICAST MTU:1500 Metric:1 RX packets:13616 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:8690528 (8.6 MB) TX bytes:0 (0.0 B) eth2 Link encap:Ethernet HWaddr 00:04:23:e1:77:6a UP BROADCAST RUNNING NOARP PROMISC ALLMULTI MULTICAST MTU:1500 Metric:1 RX packets:7750 errors:0 dropped:471 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:2780935 (2.7 MB) TX bytes:0 (0.0 B) eth3 Link encap:Ethernet HWaddr 00:04:23:e1:77:6b UP BROADCAST RUNNING NOARP PROMISC ALLMULTI MULTICAST MTU:1500 Metric:1 RX packets:5112 errors:0 dropped:206 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:639472 (639.4 KB) TX bytes:0 (0.0 B) eth4 Link encap:Ethernet HWaddr 00:04:23:b6:35:6c UP BROADCAST RUNNING NOARP PROMISC ALLMULTI MULTICAST MTU:1500 Metric:1 RX packets:961467 errors:0 dropped:935 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:958561305 (958.5 MB) TX bytes:0 (0.0 B) eth5 Link encap:Ethernet HWaddr 00:04:23:b6:35:6d inet addr:192.168.1.6 Bcast:192.168.1.255 Mask:255.255.255.0 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:4264 errors:0 dropped:16 overruns:0 frame:0 TX packets:699 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:572228 (572.2 KB) TX bytes:124456 (124.4 KB) I tried the defaults, then started to play around with settings. I wasn't using any flow control and I increased the RxDescriptor count to 4096 before the upgrade as well without any problems. # cat /etc/modprobe.d/e1000.conf options e1000 XsumRX=0,0,0,0,0 RxDescriptors=4096,4096,4096,4096,4096 FlowControl=0,0,0,0,0 debug=16 Here's my network configuration file, I turned off checksumming and various offloading mechanisms along with setting CPU affinity with heavy use interfaces getting an entire CPU and light use interfaces sharing a CPU. I used these settings prior to the upgrade without problems. # cat /etc/network/interfaces # The loopback network interface auto lo iface lo inet loopback # The primary network interface auto eth0 iface eth0 inet manual pre-up /sbin/ethtool -G eth0 rx 4096 tx 0 pre-up /sbin/ethtool -K eth0 gro off gso off rx off pre-up /sbin/ethtool -A eth0 rx off autoneg off up ifconfig eth0 0.0.0.0 -arp promisc mtu 1500 allmulti txqueuelen 0 up post-up echo "4" > /proc/irq/48/smp_affinity down ifconfig eth0 down post-down /sbin/ethtool -G eth0 rx 256 tx 256 post-down /sbin/ethtool -K eth0 gro on gso on rx on post-down /sbin/ethtool -A eth0 rx on autoneg on auto eth1 iface eth1 inet manual pre-up /sbin/ethtool -G eth1 rx 4096 tx 0 pre-up /sbin/ethtool -K eth1 gro off gso off rx off pre-up /sbin/ethtool -A eth1 rx off autoneg off up ifconfig eth1 0.0.0.0 -arp promisc mtu 1500 allmulti txqueuelen 0 up post-up echo "4" > /proc/irq/49/smp_affinity down ifconfig eth1 down post-down /sbin/ethtool -G eth1 rx 256 tx 256 post-down /sbin/ethtool -K eth1 gro on gso on rx on post-down /sbin/ethtool -A eth1 rx on autoneg on auto eth2 iface eth2 inet manual pre-up /sbin/ethtool -G eth2 rx 4096 tx 0 pre-up /sbin/ethtool -K eth2 gro off gso off rx off pre-up /sbin/ethtool -A eth2 rx off autoneg off up ifconfig eth2 0.0.0.0 -arp promisc mtu 1500 allmulti txqueuelen 0 up post-up echo "1" > /proc/irq/82/smp_affinity down ifconfig eth2 down post-down /sbin/ethtool -G eth2 rx 256 tx 256 post-down /sbin/ethtool -K eth2 gro on gso on rx on post-down /sbin/ethtool -A eth2 rx on autoneg on auto eth3 iface eth3 inet manual pre-up /sbin/ethtool -G eth3 rx 4096 tx 0 pre-up /sbin/ethtool -K eth3 gro off gso off rx off pre-up /sbin/ethtool -A eth3 rx off autoneg off up ifconfig eth3 0.0.0.0 -arp promisc mtu 1500 allmulti txqueuelen 0 up post-up echo "2" > /proc/irq/83/smp_affinity down ifconfig eth3 down post-down /sbin/ethtool -G eth3 rx 256 tx 256 post-down /sbin/ethtool -K eth3 gro on gso on rx on post-down /sbin/ethtool -A eth3 rx on autoneg on auto eth4 iface eth4 inet manual pre-up /sbin/ethtool -G eth4 rx 4096 tx 0 pre-up /sbin/ethtool -K eth4 gro off gso off rx off pre-up /sbin/ethtool -A eth4 rx off autoneg off up ifconfig eth4 0.0.0.0 -arp promisc mtu 1500 allmulti txqueuelen 0 up post-up echo "4" > /proc/irq/77/smp_affinity down ifconfig eth4 down post-down /sbin/ethtool -G eth4 rx 256 tx 256 post-down /sbin/ethtool -K eth4 gro on gso on rx on post-down /sbin/ethtool -A eth4 rx on autoneg on auto eth5 iface eth5 inet static pre-up /etc/fw.conf address 192.168.1.1 netmask 255.255.255.0 broadcast 192.168.1.255 gateway 192.168.1.1 dns-nameservers 192.168.1.2 192.168.1.3 up ifconfig eth5 up post-up echo "8" > /proc/irq/77/smp_affinity down ifconfig eth5 down Here's a few examples of packet drops, i ran one after another, probabling totaling 3 or 4 seconds. You can see increases in the drops from the 1st and 3rd. This was a non-busy time, very little traffic. # awk '{ print $1,$5 }' /proc/net/dev Inter-| face drop eth3: 225 lo: 0 eth2: 505 eth1: 0 eth5: 17 eth0: 105 eth4: 1034 # awk '{ print $1,$5 }' /proc/net/dev Inter-| face drop eth3: 225 lo: 0 eth2: 507 eth1: 0 eth5: 17 eth0: 105 eth4: 1034 # awk '{ print $1,$5 }' /proc/net/dev Inter-| face drop eth3: 227 lo: 0 eth2: 512 eth1: 0 eth5: 17 eth0: 105 eth4: 1039 I tried the pci=noacpi options. With and without, it's the same. This is what my interrupt stats looked like before the upgrade, after, with ACPI on PCI it showed multiple NICs bound to an interrupt and shared with other devices such as USB drives which I didn't like so I think i'm going to keep it with ACPI off as it's easier to designate sole purpose interrupts. Is there any advantage I would have using the default i.e. ACPI w/ PCI. ? # cat /etc/default/grub | grep CMD_LINE GRUB_CMDLINE_LINUX_DEFAULT="ipv6.disable=1 noacpi pci=noacpi" GRUB_CMDLINE_LINUX="" # cat /proc/interrupts CPU0 CPU1 CPU2 CPU3 0: 45 0 0 16 IO-APIC-edge timer 1: 1 0 0 7936 IO-APIC-edge i8042 2: 0 0 0 0 XT-PIC-XT-PIC cascade 6: 0 0 0 3 IO-APIC-edge floppy 8: 0 0 0 1 IO-APIC-edge rtc0 9: 0 0 0 0 IO-APIC-edge acpi 12: 0 0 0 1809 IO-APIC-edge i8042 14: 1 0 0 4498 IO-APIC-edge ata_piix 15: 0 0 0 0 IO-APIC-edge ata_piix 16: 0 0 0 0 IO-APIC-fasteoi uhci_hcd:usb2 18: 0 0 0 1350 IO-APIC-fasteoi uhci_hcd:usb4, radeon 19: 0 0 0 0 IO-APIC-fasteoi uhci_hcd:usb3 23: 0 0 0 4099 IO-APIC-fasteoi ehci_hcd:usb1 38: 0 0 0 61963 IO-APIC-fasteoi megaraid 48: 0 0 1002319 4 IO-APIC-fasteoi eth0 49: 0 0 38772 3 IO-APIC-fasteoi eth1 77: 0 0 130076 432159 IO-APIC-fasteoi eth4 78: 0 0 0 23917 IO-APIC-fasteoi eth5 82: 1329033 0 0 4 IO-APIC-fasteoi eth2 83: 0 4886525 0 6 IO-APIC-fasteoi eth3 NMI: 5 6 4 5 Non-maskable interrupts LOC: 61409 57076 64257 114764 Local timer interrupts SPU: 0 0 0 0 Spurious interrupts IWI: 0 0 0 0 IRQ work interrupts RES: 17956 25333 13436 14789 Rescheduling interrupts CAL: 22436 607 539 478 Function call interrupts TLB: 1525 1458 4600 4151 TLB shootdowns TRM: 0 0 0 0 Thermal event interrupts THR: 0 0 0 0 Threshold APIC interrupts MCE: 0 0 0 0 Machine check exceptions MCP: 16 16 16 16 Machine check polls ERR: 0 MIS: 0 Here's sample output of vmstat, showing the system. Barebones system right now. root@nms:~# vmstat -S m 1 procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 0 0 0 14992 192 1029 0 0 56 2 419 29 1 0 99 0 0 0 0 14992 192 1029 0 0 0 0 922 27 0 0 100 0 0 0 0 14991 192 1029 0 0 0 36 763 50 0 0 100 0 0 0 0 14991 192 1029 0 0 0 0 646 35 0 0 100 0 0 0 0 14991 192 1029 0 0 0 0 722 54 0 0 100 0 0 0 0 14991 192 1029 0 0 0 0 793 27 0 0 100 0 ^C Here's dmesg output. I can't figure out why my PCI-X slots are negotiated as PCI. The network cards are all PCI-X with the exception of the integrated NICs that came with the server. In the output below it looks as if eth3 and eth2 negotiated at PCI-X speeds rather than PCI:66Mhz. Wouldn't they all drop to PCI:66Mhz? If your integrated NICs are PCI, as labeled below (eth0,eth1), then wouldn't all devices on your bus speed drop down to that slower bus speed? If not, I still don't know why only one of my NICs ( each has two ethernet ports) is labeled as PCI-X in the output below. Does that mean it is running at PCI-X speeds are is it showing that it's capable? # dmesg | grep e1000 [ 3678.349337] e1000: Intel(R) PRO/1000 Network Driver - version 7.3.21-k8-NAPI [ 3678.349342] e1000: Copyright (c) 1999-2006 Intel Corporation. [ 3678.349394] e1000 0000:06:07.0: PCI->APIC IRQ transform: INT A -> IRQ 48 [ 3678.409725] e1000 0000:06:07.0: Receive Descriptors set to 4096 [ 3678.409730] e1000 0000:06:07.0: Checksum Offload Disabled [ 3678.409734] e1000 0000:06:07.0: Flow Control Disabled [ 3678.586409] e1000 0000:06:07.0: eth0: (PCI:66MHz:32-bit) 00:11:43:e0:e2:8c [ 3678.586419] e1000 0000:06:07.0: eth0: Intel(R) PRO/1000 Network Connection [ 3678.586642] e1000 0000:07:08.0: PCI->APIC IRQ transform: INT A -> IRQ 49 [ 3678.649854] e1000 0000:07:08.0: Receive Descriptors set to 4096 [ 3678.649859] e1000 0000:07:08.0: Checksum Offload Disabled [ 3678.649863] e1000 0000:07:08.0: Flow Control Disabled [ 3678.826436] e1000 0000:07:08.0: eth1: (PCI:66MHz:32-bit) 00:11:43:e0:e2:8d [ 3678.826444] e1000 0000:07:08.0: eth1: Intel(R) PRO/1000 Network Connection [ 3678.826627] e1000 0000:09:04.0: PCI->APIC IRQ transform: INT A -> IRQ 82 [ 3679.093266] e1000 0000:09:04.0: Receive Descriptors set to 4096 [ 3679.093271] e1000 0000:09:04.0: Checksum Offload Disabled [ 3679.093275] e1000 0000:09:04.0: Flow Control Disabled [ 3679.130239] e1000 0000:09:04.0: eth2: (PCI-X:133MHz:64-bit) 00:04:23:e1:77:6a [ 3679.130246] e1000 0000:09:04.0: eth2: Intel(R) PRO/1000 Network Connection [ 3679.130449] e1000 0000:09:04.1: PCI->APIC IRQ transform: INT B -> IRQ 83 [ 3679.397312] e1000 0000:09:04.1: Receive Descriptors set to 4096 [ 3679.397318] e1000 0000:09:04.1: Checksum Offload Disabled [ 3679.397321] e1000 0000:09:04.1: Flow Control Disabled [ 3679.434350] e1000 0000:09:04.1: eth3: (PCI-X:133MHz:64-bit) 00:04:23:e1:77:6b [ 3679.434360] e1000 0000:09:04.1: eth3: Intel(R) PRO/1000 Network Connection [ 3679.434553] e1000 0000:0a:03.0: PCI->APIC IRQ transform: INT A -> IRQ 77 [ 3679.704072] e1000 0000:0a:03.0: Receive Descriptors set to 4096 [ 3679.704077] e1000 0000:0a:03.0: Checksum Offload Disabled [ 3679.704081] e1000 0000:0a:03.0: Flow Control Disabled [ 3679.738364] e1000 0000:0a:03.0: eth4: (PCI:33MHz:64-bit) 00:04:23:b6:35:6c [ 3679.738371] e1000 0000:0a:03.0: eth4: Intel(R) PRO/1000 Network Connection [ 3679.738538] e1000 0000:0a:03.1: PCI->APIC IRQ transform: INT B -> IRQ 78 [ 3680.046060] e1000 0000:0a:03.1: eth5: (PCI:33MHz:64-bit) 00:04:23:b6:35:6d [ 3680.046067] e1000 0000:0a:03.1: eth5: Intel(R) PRO/1000 Network Connection [ 3682.132415] e1000: eth0 NIC Link is Up 100 Mbps Half Duplex, Flow Control: None [ 3682.224423] e1000: eth1 NIC Link is Up 100 Mbps Half Duplex, Flow Control: None [ 3682.316385] e1000: eth2 NIC Link is Up 100 Mbps Half Duplex, Flow Control: None [ 3682.408391] e1000: eth3 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None [ 3682.500396] e1000: eth4 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None [ 3682.708401] e1000: eth5 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX At first I thought it was the NIC drivers but I'm not so sure. I really have no idea where else to look at the moment. Any help is greatly appreciated as I'm struggling with this. If you need more information just ask. Thanks! [1]http://www.cs.fsu.edu/~baker/devices/lxr/http/source/linux/Documentation/networking/e1000.txt?v=2.6.11.8 [2] http://support.dell.com/support/edocs/systems/pe2850/en/ug/t1390aa.htm

  Read the article

< Previous Page | 1 2 3 4 5  | Next Page >