Search Results

Search found 5729 results on 230 pages for 'compiler dependent'.

Page 177/230 | < Previous Page | 173 174 175 176 177 178 179 180 181 182 183 184  | Next Page >

• Strings in ASP?

  - by TSL

  Hi All, I have some APSX code that I am trying to modify for a programmer that is out on medicaly leave. I am not an ASP guy, but rather C++ So what I want to do is delare a string, check the first 4 characters and if it is 'http' do something, if not, something else. Here is what I have: string strYT= Left(objFile, 4); if (strYT=="http") { pnlYT.Visible = true; pnlIntro.Visible = false; pnlVideo.Visible = false; } else { pnlYT.Visible = false; pnlIntro.Visible = false; pnlVideo.Visible = true; PrintText(objFile); } But I get errors like: Compiler Error Message: CS0103: The name 'Left' does not exist in the class or namespace 'ASP.zen_aspx' My googling turns up many examples of doing it just like this.....

  Read the article

• What's the outcome if I use free with new or delete with malloc?

  - by skydoor

  It is a compiler error or runtime error? The code below can be compiled! class Base{ void g(); void h(); }; int main() { Base* p = new Base(); free(p); return 0; } However it can't be compiled with a virtual function if I declare the class Base like this class Base{ virtual void g(); void h(); }; The code below can be compiled all the time, no matter the function is virtual or not. class Base{ void g(); void h(); }; int main() { Base* p = (Base*)malloc(sizeof(Base)); delete p; return 0; }

  Read the article

• C# Generics Multiple Inheritance Problem

  - by Ciemnl

  Can any one help me with this syntax issue with C#? I have no idea how to do it. class SomeClass<T> : SomeOtherClass<T> where T : ISomeInterface , IAnotherInterface { ... } I want SomeClass to inherit from SomeOtherClass and IAnotherInterface and for T to inherit ISomeInterface only It seems the problem is that the where keyword screws everything up so that the compiler thinks both ISomeInterface and IAnotherInterface should both be inherited by T. This problem is very annoying and I think the solution is some kind of parenthesis but I have tried and failed finding one that works. Also, switching around the order of the two items inherited from SomeClass does not work because the class inherited always has to come before any interfaces. I couldn't find any solutions on the MSDN C# generics pages and I can't beleive I'm the first person to have this problem. Thanks, any help is much appreciated!

  Read the article

• what is the best approach for to use openGL in the web?

  - by Y_Y

  I wrote a program in C++/OpenGL (using Dev-C++ compiler) for my calculus 2 class. The teacher liked the program and he requested me to somehow put it online so that instead of downloading the .exe file and run it the web browser will run it automatically just like a java applet. The question is: How if possible, can I display a C++/OpenGL program in a web browser? I am thinking of moving to JOGL which is a java interpretation of OpenGL but I rather stay in C++ since I am more familiar with it. Also is there any other better and easier 3D web base API that I can consider?

  Read the article

• Get result type of function

  - by Robert

  I want to specialize a template function declared as: template<typename Type> Type read(std::istream& is); I then have a lot of static implementations static int read_integer(std::istream& is); a.s.o. Now I'd like to do a macro so that specialization of read is as simple as: SPECIALIZE_READ(read_integer) So I figured I'd go the boost::function_traits way and declare SPECIALIZE_READ as: #define SPECIALIZE_READ(read_function) \ template<> boost::function_traits<read_function>::result_type read(std::istream& is) { \ return read_function(is); \ } but VC++ (2008) compiler complains with: 'boost::function_traits' : 'read_integer' is not a valid template type argument for parameter 'Function' Ideas ?

  Read the article

• XCode Project Code Changes Not Updating 100% After Save-Build-Run

  - by Greg

  When I make code changes to my iPhone game project in XCode, and then do CMD-B + Enter, I expect the project to be saved, build and run on the simulator with the latest. What is happening though, sometimes, is that it doesn't pick up a small change I make unless I clean the project and then build. I'm a long time Java person and newish to C-based languages and it's compiler. Can someone explain to me what is cached after each build that does this and how to change my project settings to avoid having to clean everytime? Or tell me the bad news that this is part of C development? Not trying to bash it - I get compiled JSPs stuck in the working cache often in Java, too. :P

  Read the article

• C++ template + typedef

  - by MMS

  What is wrong in the following code: Point2D.h template <class T> class Point2D { private: T x; T y; ... }; PointsList.h template <class T> class Point2D; template <class T> struct TPointsList { typedef std::vector <Point2D <T> > Type; }; template <class T> class PointsList { private: TPointsList <T>::Type points; //Compiler error ... }; I would like to create new user type TPointsList without direct type specification...

  Read the article

• Tracking Useful Information

  - by Steve M

  What do the clever programmers here do to keep track of handy programming tricks and useful information they pick up over their many years of experience? Things like useful compiler arguments, IDE short-cuts, clever code snippets, etc. I sometimes find myself frustrated when looking up something that I used to know a year or two ago. My IE favourites probably represent a good chunk of the Internet in the late 1990s, so clearly that isn't effective (at least for me). Or am I just getting old? So.. what do you do?

  Read the article

• Ruby On Rails -

  - by Adam S

  I am trying to create a collection_select that is dependent on another collection_select, following the railscasts episode #88.. The database includes a schedule of all cruise ships arrival dates. See attached schema diagram,... (The arrow side of the lines indicate has_many, the non-pointy side indicates belongs_to) In the "booking/new" view I have a collection_select for choosing a cruiseline, then another collection_select will appear for selecting a ship, then another for selecting the date that ship is in. If I ONLY put a collection_select for shipshedule it works fine(because of the direct association with the bookings model). However, if I try to add a collection_select for cruiselines....it breaks(Im assuming because their isnt a direct association). The collection_select for cruiselines returns an undefined method error for "cruiseline_id".....if I simply use "id" the collection_select works, but of course isnt fully functional due to incorrect naming of the form field. I have tried "has_many :shipschedules, :through = :cruiseships" in the cruiseline model and "has_many :bookings, :through = :shipschedules" in the cruiseships model... As you can see in the diagram, I need to access the cruiseships and cruiselines model through the bookings model. I have the models set up so a cruiseline has_many cruiseships, a cruiseship has_many shipschedules, and a shipschedule has_many bookings. But, the bookings model cant directly access the cruiseline model. How do I accomplish this? THANKS!!! http://www.adamstockland.com/common/images/RoRf/PastedGraphic-2.png http://www.adamstockland.com/common/images/RoRf/PastedGraphic-1.png

  Read the article

• question about copy constructor

  - by lego69

  I have this class: class A { private: int player; public: A(int initPlayer = 0); A(const A&); A& operator=(const A&); ~A(); void foo() const; }; and I have function which contains this row: A *pa1 = new A(a2); can somebody please explain what exactly is going on, when I call A(a2) compiler calls copy constructor or constructor, thanks in advance

  Read the article

• Reading a directory

  - by paleman

  Hi, I'm trying to solve exercise from K&R, it's about reading directories.This task is system dependent because it uses system calls.In the book example authors say that their example is written for Version 7 and System V UNIX systems and that they used the directory information in the header < sys/dir.h,which looks like this: #ifndef DIRSIZ #define DIRSIZ 14 #endif struct direct { /* directory entry */ ino_t d_ino; /* inode number */ char d_name[DIRSIZ]; /* long name does not have '\0' */ }; On this system they use 'struct direct' combined with 'read' function to retrieve a directory entry, which consist of file name and inode number. ..... struct direct dirbuf; /* local directory structure */ while(read(dp->fd, (char *) &dirbuf, sizeof(dirbuf) == sizeof(dirbuf) { ..... } ..... I suppose this works fine on UNIX and Linux systems, but what I want to do is modify this so it works on Windows XP. Is there some structure in Windows like 'struct direct' so I can use it with 'read' function and if there is what is the header name where it is defined? Or maybe Windows requires completely different approach?

  Read the article

• Pointer aliasing- in C++0x

  - by DeadMG

  I'm thinking about (just as an idea) disjointed pointer aliasing in C++0x. I was thinking about seeing if it could be implemented similarly to const correctness- that is, enforced by the compiler. What would be the requirements for such a thing? As this is more of a thought experiment, I'm perfectly happy to look at solutions that destroy legacy code or redefine half the language and that kind of thing. What I'd really rather not do is have, say, restrict from C99 where the programmer just promises it. It should be enforced.

  Read the article

• How to make Eclipse compile with Java 1.5 in Mac OS X Leopard (10.5)?

  - by whiskeyspider

  I have been developing in Snow Leopard (10.6). I believe I have some Java6-only API features and I'd like to find these. I moved my project to Leopard (10.5) -- since Snow Leopard does not support Java 1.5 -- but I am so far unable to make Eclipse give compile errors for a Java API call which I know to have been added in 1.6. Here is what I have done: Project properties - Java Compiler - set everything to 1.5. Preferences - Java - Installed JREs - selected JVM 1.5.0. Project properties - Java Build Path - Libraries - says JRE System Library [JVM 1.5.0] What am I missing? Is there another way to find Java API calls added in 1.6 when using Eclipse in Snow Leopard? Thanks.

  Read the article

• Can we overload a function based on only whether a parameter is a value or a reference?

  - by skydoor

  I got the answer NO! Because passing by value and passing by reference looks identical to the caller. However, the code below compiles right class A { public: void f(int i) {} void f(int& i) {} }; But when I try to use it, there is compile error. int main () { A a; int i = 9; int& j = i; a.f(1); a.f(i); a.f(j); return 0; } Why does not the compiler disable it even without knowing it is going to be used?

  Read the article

• Testing IPhone code on Windows

  - by steve

  I'm picking up a new Dell laptop. My primary machine is a IMac. I will most likely have to write some IPhone projects for someone in the future. While I do most of my work on the IMac there would be maybe 25% of the time where I work from my laptop. Can anyone tell me if I use objective C / IPhone SDK's if there is a generic objective C compiler I can use to see if my code would in theroy work? Not looking to do hackintosh or anything like that. My other option is to just get a discounted mac mini (Think this is most likely) as well as the Dell. Thanks for any advice

  Read the article

• Generics not so generic !!

  - by Aymen

  Hi I tried to implement a generic binary search algorithm in scala. Here it is : type Ord ={ def <(x:Any):Boolean def >(x:Any):Boolean } def binSearch[T <: Ord ](x:T,start:Int,end:Int,t:Array[T]):Boolean = { if (start > end) return false val pos = (start + end ) / 2 if(t(pos)==x) true else if (t(pos) < x) binSearch(x,pos+1,end,t) else binSearch(x,start,pos-1,t) } everything is OK until I tried to actually use it (xD) : binSearch(3,0,4,Array(1,2,5,6)) the compiler is pretending that Int not a member of Ord, well what shall I do to solve this ? Thanks

  Read the article

• @MustOverride annotation?

  - by Harrypotter2k5

  In .NET, one can specify a "mustoverride" attribute to a method in a particular superclass to ensure that subclasses override that particular method. I was wondering whether anybody has a custom java annotation that could achieve the same effect. Essentially what i want is to push for subclasses to override a method in a superclass that itself has some logic that must be run-through. I dont want to use abstract methods or interfaces, because i want some common functionality to be run in the super method, but more-or-less produce a compiler warning/error denoting that derivative classes should override a given method.

  Read the article

• Returning a struct from a class method

  - by tree

  I have a header file that looks something like the following: class Model { private: struct coord { int x; int y; } xy; public: .... coord get() const { return xy; } }; And in yet another file (assume ModelObject exists): struct c { int x; int y; void operator = (c &rhs) { x = rhs.x; y = rhs.y; }; } xy; xy = ModelObject->get(); The compiler throws an error that says there is no known covnersion from coord to c. I believe it is because it doesn't know about coord type because it is declared inside of a class header. I can get around that by declaring the struct outside of the class, but I was wondering if it is possible to do the way I am, or is this generally considered bad practice

  Read the article

• Java Basics: create class object

  - by user1767853

  In C++: class Rectangle { int x, y; public: void set_values (int,int); int area () {return (x*y);} }; int main () { Rectangle rect; rect.set_values (3,4); } In Java: class Rectangle { int x, y; void set_values (int x,int y); int area () {return (x*y);} } public static void main(String[] args) { Rectangle rect=new Rectangle(3,4); } In C++ compiler will create rect object & reserve memory 4 bytes. I want to know How Java is creating object?

  Read the article

• printf("... %c ...",'\0') and family - what will happen?

  - by SF.

  How will various functions that take printf format string behave upon encountering the %c format given value of \0/NULL? How should they behave? Is it safe? Is it defined? Is it compiler-specific? e.g. sprintf() - will it crop the result string at the NULL? What length will it return? Will printf() output the whole format string or just up to the new NULL? Will va_args + vsprintf/vprintf be affected somehow? If so, how? Do I risk memory leaks or other problems if I e.g. shoot this NULL at a point in std::string.c_str()? What are the best ways to avoid this caveat (sanitize input?)

  Read the article

• Multiple subnets on isc-dhcp-server using ddns with bind9

  - by legioxi

  On my network I have two subnets: 10.100.1.0/24 - Wired/wireless 10.100.7.0/24 - VPN Both subnets are served by isc-dhcp-server running on a Debian VM. This same VM runs bind9 for my DNS. ISC-DHCP-SERVER is configured to use DDNS and update BIND9 with hosts/IPs. Everything runs great until a device drops off the wired/wireless network and pops onto the VPN. When connecting on the VPN, a DHCP lease is handed out on the new subnet but DDNS does not update BIND9. Since the device has A/TXT/PTR records it appears ISC-DHCP-SERVER won't switch them to the new IP. The logs show: Connect to wireless: Nov 6 20:55:13 core-server named[2417]: client 127.0.0.1#57697: updating zone 'internal.mydomain.com/IN': adding an RR at 'demo-iphone.internal.mydomain.com' A Nov 6 20:55:13 core-server named[2417]: client 127.0.0.1#57697: updating zone 'internal.mydomain.com/IN': adding an RR at 'demo-iphone.internal.mydomain.com' TXT Nov 6 20:55:13 core-server dhcpd: DHCPACK on 10.100.1.160 to FF:FF:FF:FF:FF:FF (demo-iphone) via eth0 Nov 6 20:55:13 core-server dhcpd: Added new forward map from demo-iphone.internal.mydomain.com to 10.100.1.160 Nov 6 20:55:13 core-server dhcpd: Added reverse map from 160.49.21.172.in-addr.arpa. to demo-iphone.internal.mydomain.com Switch to VPN: Nov 6 20:56:34 core-server dhcpd: DHCPOFFER on 10.100.7.101 to BB:BB:BB:BB:BB:BB (demo-iphone) via 10.100.7.0 Nov 6 20:56:34 core-server named[2417]: client 127.0.0.1#57697: updating zone 'internal.mydomain.com/IN': update unsuccessful: demo-iphone.internal.mydomain.com: 'name not in use' prerequisite not satisfied (YXDOMAIN) Nov 6 20:56:34 core-server dhcpd: DHCPREQUEST for 10.100.7.101 (10.100.1.2) from BB:BB:BB:BB:BB:BB (demo-iphone) via 10.100.7.0 Nov 6 20:56:34 core-server dhcpd: DHCPACK on 10.100.7.101 to BB:BB:BB:BB:BB:BB (demo-iphone) via 10.100.7.0 Nov 6 20:56:34 core-server named[2417]: client 127.0.0.1#57697: updating zone 'internal.mydomain.com/IN': update unsuccessful: demo-iphone.internal.mydomain.com/TXT: 'RRset exists (value dependent)' prerequisite not satisfied (NXRRSET) Nov 6 20:56:34 core-server dhcpd: Forward map from demo-iphone.internal.mydomain.com to 10.100.7.101 FAILED: Has an address record but no DHCID, not mine. One thing to note is that the MAC of the device when connecting via VPN is the MAC of my Cisco ASA5512X and not the actual device. The ASA is relaying the DHCP request from the VPN client to the VM running ISC-DHCP-SERVER. Is there a way to get DDNS working in this scenario?

  Read the article

• HP UX can not boot from Ignite Tape

  - by Spirit

  We have hp rp2470 server running hp-ux 11.00, with LVM mirroring. As for redundancy we have second rp2470 same hw (same two processors, same ram, same two hdd’s, same number of lan cards). I want to clone first one to the second. For that purpose I am making ignite tape with the following command: make_tape_recovery -x inc_entire=vg00 Ignite tape finishes without problems. When I boot second server from this ignate tape, server is starting to boot, and ignite restore finishes without any errors, only few notes, which are normal. However vmunix is not booting and when restore finishes, it boot to ISL prompt. From this I cannot boot /stand/vmunix. I tried to run recovery shell but no success. When recovery shell ask to do frecover to restore critical files, then I receive error: frecover(5405): unable to open /dev/rmt/0m At first I thought that the problem might be in the difference of the firmware version of the servers: fw version of production server is: Firmware Version 43.50 and fw version of backup server is: Firmware Version 42.19 So i did a fw upgrade of my backup server so that both servers are v43.50, and tried a recovery but again cant boot the system. Next I did another archive tape with -I (Interactive) flag: make_tape_recovery -I -x inc_entire=vg00 and tried recovery with it, again no good. I cannot find any error or warnings on ignite log, and I cannot boot hpux. I am only on ISL prompt. This is what i've noticed on the gsp logs: ************* SYSTEM ALERT ************** SYSTEM NAME: mcnfwim1 DATE: 07/27/2003 TIME: 10:18:49 ALERT LEVEL: 6 = Boot possible, pending failure - action required REASON FOR ALERT SOURCE: 8 = I/O SOURCE DETAIL: 6 = disk SOURCE ID: 0 PROBLEM DETAIL: 0 = no problem detail LEDs: RUN ATTENTION FAULT REMOTE POWER FLASH OFF ON ON ON LED State: Boot Failed. Running non-OS code. Check Chassis and Console Logs for error messages. 0x00000060860010B0 00000000 00000000 - type 0 = Data Field Unused 0x58000860860010B0 00006706 1B0A1231 - type 11 = Timestamp 07/27/2003 10:18:49 And another gsp log: Log Entry # 3 : SYSTEM NAME: mcnfwim1 DATE: 07/27/2003 TIME: 10:12:20 ALERT LEVEL: 6 = Boot possible, pending failure - action required SOURCE: 8 = I/O SOURCE DETAIL: 6 = disk SOURCE ID: 0 PROBLEM DETAIL: 0 = no problem detail CALLER ACTIVITY: 1 = test STATUS: 0 CALLER SUBACTIVITY: 0B = implementation dependent REPORTING ENTITY TYPE: 0 = system firmware REPORTING ENTITY ID: 00 0x00000060860010B0 00000000 00000000 type 0 = Data Field Unused 0x58000860860010B0 00006706 1B0A0C14 type 11 = Timestamp 07/27/2003 10:12:20 Type CR for next entry, - CR for previous entry, Q CR to quit. Please note that I can not change anything on the production server. I can only make changes to the backup server. Any help is appreciated.

  Read the article

• How to get physical partition name from iSCSI details on Windows?

  - by Barry Kelly

  I've got a piece of software that needs the name of a partition in \Device\Harddisk2\Partition1 style, as shown e.g. in WinObj. I want to get this partition name from details of the iSCSI connection that underlies the partition. The trouble is that disk order is not fixed - depending on what devices are connected and initialized in what order, it can move around. So suppose I have the portal name (DNS of the iSCSI target), target IQN, etc. I'd like to somehow discover which volumes in the system relate to it, in an automated fashion. I can write some PowerShell WMI queries that get somewhat close to the desired info: PS> get-wmiobject -class Win32_DiskPartition NumberOfBlocks : 204800 BootPartition : True Name : Disk #0, Partition #0 PrimaryPartition : True Size : 104857600 Index : 0 ... From the Name here, I think I can fabricate the corresponding name by adding 1 to the partition number: \Device\Harddisk0\Partition1 - Partition0 appears to be a fake partition mapping to the whole disk. But the above doesn't have enough information to map to the underlying physical device, unless I take a guess based on exact size matching. I can get some info on SCSI devices, but it's not helpful in joining things up (iSCSI target is Nexenta/Solaris COMSTAR): PS> get-wmiobject -class Win32_SCSIControllerDevice __GENUS : 2 __CLASS : Win32_SCSIControllerDevice ... Antecedent : \\COBRA\root\cimv2:Win32_SCSIController.DeviceID="ROOT\\ISCSIPRT\\0000" Dependent : \\COBRA\root\cimv2:Win32_PnPEntity.DeviceID="SCSI\\DISK&VEN_NEXENTA&PROD_COMSTAR... Similarly, I can run queries like these: PS> get-wmiobject -namespace ROOT\WMI -class MSiSCSIInitiator_TargetClass PS> get-wmiobject -namespace ROOT\WMI -class MSiSCSIInitiator_PersistentDevices These guys return information relating to my iSCSI target name and the GUID volume name respectively (a volume name like \\?\Volume{guid-goes-here}), but the GUID volume name is no good to me, and there doesn't appear to be a reliable correspondence between the target name and the volume that I can join on. I simply can't find an easy way of getting from an IQN (e.g. iqn.1992-01.com.example:storage:diskarrays-sn-a8675309) to physical partitions mapped from that target. The way I do it by hand? I start Disk Management, and look for a partition of the correct size, verify that its driver says NEXENTA COMSTAR, and look at the disk number. But even this is unreliable if I have multiple iSCSI volumes of the exact same size. Any suggestions?

  Read the article

• How to keep group-writeable shares on Samba with OSX clients?

  - by Oliver Salzburg

  I have a FreeNAS server on a network with OSX and Windows clients. When the OSX clients interact with SMB/CIFS shares on the server, they are causing permission problems for all other clients. Update: I can no longer verify any answers because we abandoned the project, but feel free to post any help for future visitors. The details of this behavior seem to also be dependent on the version of OSX the client is running. For this question, let's assume a client running 10.8.2. When I mount the CIFS share on an OSX client and create a new directory on it, the directory will be created with drwxr-x-rx permissions. This is undesirable because it will not allow anyone but me to write to the directory. There are other users in my group which should have write permissions as well. This behavior happens even though the following settings are present in smb.conf on the server: [global] create mask= 0666 directory mask= 0777 [share] force directory mode= 0775 force create mode= 0660 I was under the impression that these settings should make sure that directories are at least created with rwxrwxr-x permissions. But, I guess, that doesn't stop the client from changing the permissions after creating the directory. When I create a folder on the same share from a Windows client, the new folder will have the desired access permissions (rwxrwxrwx), so I'm currently assuming that the problem lies with the OSX client. I guess this wouldn't be such an issue if you could easily change the permissions of the directories you've created, but you can't. When opening the directory info in Finder, I get the old "You have custom access" notice with no ability to make any changes. I'm assuming that this is caused because we're using Windows ACLs on the share, but that's just a wild guess. Changing the write permissions for the group through the terminal works fine, but this is unpractical for the deployment and unreasonable to expect from anyone to do. This is the complete smb.conf: [global] encrypt passwords = yes dns proxy = no strict locking = no read raw = yes write raw = yes oplocks = yes max xmit = 65535 deadtime = 15 display charset = LOCALE max log size = 10 syslog only = yes syslog = 1 load printers = no printing = bsd printcap name = /dev/null disable spoolss = yes smb passwd file = /var/etc/private/smbpasswd private dir = /var/etc/private getwd cache = yes guest account = nobody map to guest = Bad Password obey pam restrictions = Yes # NOTE: read smb.conf. directory name cache size = 0 max protocol = SMB2 netbios name = freenas workgroup = COMPANY server string = FreeNAS Server store dos attributes = yes hostname lookups = yes security = user passdb backend = ldapsam:ldap://ldap.company.local ldap admin dn = cn=admin,dc=company,dc=local ldap suffix = dc=company,dc=local ldap user suffix = ou=Users ldap group suffix = ou=Groups ldap machine suffix = ou=Computers ldap ssl = off ldap replication sleep = 1000 ldap passwd sync = yes #ldap debug level = 1 #ldap debug threshold = 1 ldapsam:trusted = yes idmap uid = 10000-39999 idmap gid = 10000-39999 create mask = 0666 directory mask = 0777 client ntlmv2 auth = yes dos charset = CP437 unix charset = UTF-8 log level = 1 [share] path = /mnt/zfs0 printable = no veto files = /.snap/.windows/.zfs/ writeable = yes browseable = yes inherit owner = no inherit permissions = no vfs objects = zfsacl guest ok = no inherit acls = Yes map archive = No map readonly = no nfs4:mode = special nfs4:acedup = merge nfs4:chown = yes hide dot files force directory mode = 0775 force create mode = 0660

  Read the article

• MongoDB and datasets that don't fit in RAM no matter how hard you shove

  - by sysadmin1138

  This is very system dependent, but chances are near certain we'll scale past some arbitrary cliff and get into Real Trouble. I'm curious what kind of rules-of-thumb exist for a good RAM to Disk-space ratio. We're planning our next round of systems, and need to make some choices regarding RAM, SSDs, and how much of each the new nodes will get. But now for some performance details! During normal workflow of a single project-run, MongoDB is hit with a very high percentage of writes (70-80%). Once the second stage of the processing pipeline hits, it's extremely high read as it needs to deduplicate records identified in the first half of processing. This is the workflow for which "keep your working set in RAM" is made for, and we're designing around that assumption. The entire dataset is continually hit with random queries from end-user derived sources; though the frequency is irregular, the size is usually pretty small (groups of 10 documents). Since this is user-facing, the replies need to be under the "bored-now" threshold of 3 seconds. This access pattern is much less likely to be in cache, so will be very likely to incur disk hits. A secondary processing workflow is high read of previous processing runs that may be days, weeks, or even months old, and is run infrequently but still needs to be zippy. Up to 100% of the documents in the previous processing run will be accessed. No amount of cache-warming can help with this, I suspect. Finished document sizes vary widely, but the median size is about 8K. The high-read portion of the normal project processing strongly suggests the use of Replicas to help distribute the Read traffic. I have read elsewhere that a 1:10 RAM-GB to HD-GB is a good rule-of-thumb for slow disks, As we are seriously considering using much faster SSDs, I'd like to know if there is a similar rule of thumb for fast disks. I know we're using Mongo in a way where cache-everything really isn't going to fly, which is why I'm looking at ways to engineer a system that can survive such usage. The entire dataset will likely be most of a TB within half a year and keep growing.

  Read the article

< Previous Page | 173 174 175 176 177 178 179 180 181 182 183 184  | Next Page >