Search Results

Search found 9634 results on 386 pages for 'proxy pattern'.

Page 316/386 | < Previous Page | 312 313 314 315 316 317 318 319 320 321 322 323  | Next Page >

  • maximum number of connections Squid

    - by Isaac
    I have a Squid proxy server that controls all internet traffic for my network. I need a way to stop users from downloading big files (say 50MB) in my network. I banned some famous ports (e.g. torrent) but some downloads are possible by HTTP port. Obviously I cannot ban port 80! A simple solution is limiting maxmimum number of the simultaneous connections for each IP (e.g. 3 connections). It's possible in Squid with this config: acl ACCOUNTSDEPT 192.168.5.0/24 acl limitusercon maxconn 3 http_access deny ACCOUNTSDEPT limitusercon But this solution has really bad impact in web browsing, because any smart browser get different parts of a website by several connections simultaneously to speedup web browsing. But if we have a maximum number of connections, the browsers will fail to get some parts and the website will be shown partially and some parts/images/frames will not be shown. So, can we limit maximum number of persist connections? I think this policy will works: Specify Maximum number of connections that is alive for 10 seconds But Number of simultaneous connections for every IP is unlimited But how can we implement this policy when Squid? With which config? UPDATE: artifex and Tom Newton offered using a bandwidth-limiting approach to fight against downloaders. But bandwidth-limiting in Squid has a shortcoming: It's static and cannot dynamically change. So a person has a limited bandwidth not matter how many people are using internet (maybe nobody!) Also, this solution cannot help to stop people from downloading. They still can download but in a lower speed. But if we find a way to terminate persist connections (or any connection that is alive more than a specific time), downloading big files will be almost impossible (always there is some way!)

    Read the article

  • What is auto-mounting my media volume?

    - by user285277
    Something is repeatedly mounting my "media" share, doing something with it, then quietly un-mounting it with no notifications at the user level. from the little I can gleaned from the console messages below, I thought I'd managed to stop it, if not understand it last night when I followed instructions for deleting all traces of the Google Update Daemon. I've not been using any Google apps whatsoever, so I was surprised to see that in Console. What's more surprising, and perhaps a little distressing, is that the same thing occurred this evening, when the Google Daemon is long gone. I don't have that log because I can't recall precisely what time it occurred. I'll do a search and provide it if you wish, though. In the meantime, any help with this would be extremely appreciated. I've asked over at Apple Discussions but I think it might be a little deeper than those manning the boards this weekend are comfortable with. It's certainly beyond my meager skills. With apologies in advance if this is more lines thank you need. Please note that I've transformed the Google links a little because the forum here requires more reputation points before one can post more than two links. 12/27/13 10:47:31.000 PM kernel[0]: memorystatus_thread: idle exiting pid 53629 [distnoted] 12/27/13 10:48:10.433 PM com.apple.Preview.TrustedBookmarksService[53640]: Failed to resolve bookmark data at index: 0; not stale; error: The file doesn’t exist. 12/27/13 10:48:10.434 PM com.apple.Preview.TrustedBookmarksService[53640]: Failed to resolve bookmark data at index: 1; not stale; error: The file doesn’t exist. 12/27/13 10:48:10.950 PM com.apple.SecurityServer[17]: Session 103257 created 12/27/13 10:48:34.328 PM com.apple.Preview.TrustedBookmarksService[53640]: Failed to resolve bookmark data at index: 2; not stale; error: The file doesn’t exist. 12/27/13 10:48:34.000 PM kernel[0]: AFP_VFS afpfs_mount: /Volumes/Media Archive-1, pid 53641 12/27/13 10:48:34.000 PM kernel[0]: AFP_VFS afpfs_mount : succeeded on volume 0xffffff80d6355008 /Volumes/Media Archive-1 (error = 0, retval = 0) 12/27/13 10:49:32.000 PM kernel[0]: wlEvent: en0 en0 Link DOWN virtIf = 0 12/27/13 10:49:32.000 PM kernel[0]: AirPort: Link Down on en0. Reason 8 (Disassociated because station leaving). 12/27/13 10:49:32.000 PM kernel[0]: en0::IO80211Interface::postMessage bssid changed 12/27/13 10:49:33.681 PM configd[16]: network changed: v4(en0-:10.0.1.12) DNS- Proxy- SMB 12/27/13 10:49:33.697 PM configd[16]: network changed: DNS* Proxy 12/27/13 10:49:35.475 PM KernelEventAgent[57]: tid 00000000 received event(s) VQ_NOTRESP (1) 12/27/13 10:49:35.000 PM kernel[0]: ASP_TCP Disconnect: triggering reconnect by bumping reconnTrigger from curr value 0 on so 0xffffff802176b4a0 12/27/13 10:49:35.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect started /Volumes/Media Archive-1 prevTrigger 0 currTrigger 1 12/27/13 10:49:35.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: doing reconnect on /Volumes/Media Archive-1 12/27/13 10:49:35.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: posting to KEA EINPROGRESS for /Volumes/Media Archive-1 12/27/13 10:49:35.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: Max reconnect time: 600 secs, Connect timeout: 15 secs for /Volumes/Media Archive-1 12/27/13 10:49:35.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: connect to the server /Volumes/Media Archive-1 12/27/13 10:49:35.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: connect on /Volumes/Media Archive-1 failed 65. 12/27/13 10:49:35.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: sleep for 1 seconds and then try again 12/27/13 10:49:35.479 PM KernelEventAgent[57]: tid 00000000 type 'afpfs', mounted on '/Volumes/Media Archive-1', from '//Me@Capsule._afpovertcp._tcp.local/Media%20Archive', not responding 12/27/13 10:49:35.487 PM KernelEventAgent[57]: tid 00000000 found 1 filesystem(s) with problem(s) 12/27/13 10:49:36.692 PM com.bourgeoisbits.cloak.agent[14503]: NetworkProfile: (null), (null), (null) (Connected: NO, Airport: NO, Open: NO) [trusted] 12/27/13 10:49:36.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: connect to the server /Volumes/Media Archive-1 12/27/13 10:49:36.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: connect on /Volumes/Media Archive-1 failed 65. 12/27/13 10:49:36.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: sleep for 2 seconds and then try again 12/27/13 10:49:38.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: connect to the server /Volumes/Media Archive-1 12/27/13 10:49:38.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: connect on /Volumes/Media Archive-1 failed 65. 12/27/13 10:49:38.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: sleep for 4 seconds and then try again 12/27/13 10:49:41.000 PM kernel[0]: CODE SIGNING: cs_invalid_page(0x1000): p=53662[GoogleSoftwareUp] clearing CS_VALID 12/27/13 10:49:42.102 PM GoogleSoftwareUpdateDaemon[53663]: -[KeystoneDaemon logServiceState] GoogleSoftwareUpdate daemon (1.1.0.3659) vending: com.google.Keystone.Daemon.UpdateEngine: 2 connection(s) com.google.Keystone.Daemon.Administration: 0 connection(s) 12/27/13 10:49:42.113 PM GoogleSoftwareUpdateDaemon[53663]: -[KSUpdateEngine updateProductID:] KSUpdateEngine updating product ID: "com.google.Keystone" 12/27/13 10:49:42.116 PM GoogleSoftwareUpdateDaemon[53663]: -[KSCheckAction performAction] KSCheckAction checking 1 ticket(s). 12/27/13 10:49:42.121 PM GoogleSoftwareUpdateDaemon[53663]: -[KSUpdateCheckAction performAction] KSUpdateCheckAction starting update check for ticket(s): {( <KSTicket:0x531870 productID=com.google.Keystone version=1.1.0.3659 xc=<KSPathExistenceChecker:0x5302d0 path=/Library/Google/GoogleSoftwareUpdate/GoogleSoftwareUpdate.bundle/> serverType=Omaha url=htt[PeeEs]://tools.google.com/service/update2 creationDate=2012-08-12 14:47:10 > )} Using server: <KSOmahaServer:0x534340 engine=<KSDaemonUpdateEngine:0x52e530> params={ EngineVersion = "1.1.0.3659"; ActivesInfo = { "com.google.talkplugin" = { LastRollCallPingDate = 2013-10-06 07:00:00 +0000; }; "com.google.Keystone" = { LastRollCallPingDate = 2013-10-06 07:00:00 +0000; LastActivePingDate = 2013-10-06 07:00:00 +0000; LastActiveDate = 2013-12-28 03:49:42 +0000; }; "com.google.picasa" = { LastActiveDate = 2012-08-29 10:15:42 +0000; }; }; UserInitiated = 0; IsSystem = 1; OmahaOSVersion = "10.8.5_i486"; Identity = KeystoneDaemon; AllowedSubdomains = ( ".omaha.sandbox.google.com", ".tools.google.com", ".www.google.com", ".corp.google.com" ); } > 12/27/13 10:49:42.130 PM GoogleSoftwareUpdateDaemon[53663]: -[KSUpdateCheckAction performAction] KSUpdateCheckAction running KSServerUpdateRequest: <KSOmahaServerUpdateRequest:0x1a31a90 server=<KSOmahaServer:0x534340> url="htt[PeeEs]://tools.google.com/service/update2" runningFetchers=0 tickets=1 activeTickets=1 rollCallTickets=1 body= <?xml version="1.0" encoding="UTF-8" standalone="yes"?> <o:gupdate xmlns:o="htt[Pee]://www.google.com/update2/request" protocol="2.0" version="KeystoneDaemon-1.1.0.3659" ismachine="1"> <o:os platform="mac" version="MacOSX" sp="10.8.5_i486"></o:os> <o:app appid="com.google.Keystone" version="1.1.0.3659" lang="en-us" installage="502" brand="GGLG"> <o:ping r="82" a="82"></o:ping> <o:updatecheck></o:updatecheck> </o:app> </o:gupdate> > 12/27/13 10:49:42.291 PM GoogleSoftwareUpdateDaemon[53663]: -[KSOutOfProcessFetcher(PrivateMethods) helperDidTerminate:] The Internet connection appears to be offline. [NSURLErrorDomain:-1009] 12/27/13 10:49:42.291 PM GoogleSoftwareUpdateDaemon[53663]: -[KSServerUpdateRequest(PrivateMethods) fetcher:failedWithError:] KSServerUpdateRequest fetch failed. (productIDs: com.google.Keystone) [com.google.UpdateEngine.CoreErrorDomain:702 - 'htt[PeeEs]://tools.google.com/service/update2'] (The Internet connection appears to be offline. [NSURLErrorDomain:-1009]) 12/27/13 10:49:42.292 PM GoogleSoftwareUpdateDaemon[53663]: -[KSUpdateCheckAction(PrivateMethods) finishAction] KSUpdateCheckAction found updates: {( )} 12/27/13 10:49:42.295 PM GoogleSoftwareUpdateDaemon[53663]: -[KSPrefetchAction performAction] KSPrefetchAction no updates to prefetch. 12/27/13 10:49:42.295 PM GoogleSoftwareUpdateDaemon[53663]: -[KSMultiUpdateAction performAction] KSSilentUpdateAction had no updates to apply. 12/27/13 10:49:42.296 PM GoogleSoftwareUpdateDaemon[53663]: -[KSMultiUpdateAction performAction] KSPromptAction had no updates to apply. 12/27/13 10:49:42.299 PM GoogleSoftwareUpdateDaemon[53663]: -[KSUpdateEngine(PrivateMethods) updateFinish] KSUpdateEngine update processing complete. 12/27/13 10:49:42.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: connect to the server /Volumes/Media Archive-1 12/27/13 10:49:42.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: connect on /Volumes/Media Archive-1 failed 65. 12/27/13 10:49:42.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: sleep for 8 seconds and then try again 12/27/13 10:49:43.152 PM GoogleSoftwareUpdateDaemon[53663]: -[KSUpdateEngine updateAllProducts] KSUpdateEngine updating all installed products. 12/27/13 10:49:43.153 PM GoogleSoftwareUpdateDaemon[53663]: -[KSCheckAction performAction] KSCheckAction checking 2 ticket(s). 12/27/13 10:49:43.158 PM GoogleSoftwareUpdateDaemon[53663]: -[KSUpdateCheckAction performAction] KSUpdateCheckAction starting update check for ticket(s): {( <KSTicket:0x18367a0 productID=com.google.Keystone version=1.1.0.3659 xc=<KSPathExistenceChecker:0x1837e10 path=/Library/Google/GoogleSoftwareUpdate/GoogleSoftwareUpdate.bundle/> serverType=Omaha url=htt[PeeEs]://tools.google.com/service/update2 creationDate=2012-08-12 14:47:10 >, <KSTicket:0x1834750 productID=com.google.talkplugin version=4.7.0.15362 xc=<KSPathExistenceChecker:0x1833890 path=/Library/Application Support/Google/GoogleTalkPlugin.app> serverType=Omaha url=htt[PeeEs]://tools.google.com/service/update2 creationDate=2012-08-12 14:47:10 > )} Using server: <KSOmahaServer:0x52e930 engine=<KSDaemonUpdateEngine:0x52e530> params={ EngineVersion = "1.1.0.3659"; ActivesInfo = { "com.google.talkplugin" = { LastRollCallPingDate = 2013-10-06 07:00:00 +0000; }; "com.google.Keystone" = { LastRollCallPingDate = 2013-10-06 07:00:00 +0000; LastActivePingDate = 2013-10-06 07:00:00 +0000; LastActiveDate = 2013-12-28 03:49:42 +0000; }; "com.google.picasa" = { LastActiveDate = 2012-08-29 10:15:42 +0000; }; }; UserInitiated = 0; IsSystem = 1; OmahaOSVersion = "10.8.5_i486"; Identity = KeystoneDaemon; AllowedSubdomains = ( ".omaha.sandbox.google.com", ".tools.google.com", ".www.google.com", ".corp.google.com" ); } > 12/27/13 10:49:43.159 PM GoogleSoftwareUpdateDaemon[53663]: -[KSUpdateCheckAction performAction] KSUpdateCheckAction running KSServerUpdateRequest: <KSOmahaServerUpdateRequest:0x53a210 server=<KSOmahaServer:0x52e930> url="htt[PeeEs]://tools.google.com/service/update2" runningFetchers=0 tickets=2 activeTickets=1 rollCallTickets=2 body= <?xml version="1.0" encoding="UTF-8" standalone="yes"?> <o:gupdate xmlns:o="htt[Pee]://www.google.com/update2/request" protocol="2.0" version="KeystoneDaemon-1.1.0.3659" ismachine="1"> <o:os platform="mac" version="MacOSX" sp="10.8.5_i486"></o:os> <o:app appid="com.google.Keystone" version="1.1.0.3659" lang="en-us" installage="502" brand="GGLG"> <o:ping r="82" a="82"></o:ping> <o:updatecheck></o:updatecheck> </o:app> <o:app appid="com.google.talkplugin" version="4.7.0.15362" lang="en-us" installage="502" brand="GGLG"> <o:ping r="82"></o:ping> <o:updatecheck></o:updatecheck> </o:app> </o:gupdate> > 12/27/13 10:49:43.243 PM GoogleSoftwareUpdateDaemon[53663]: -[KSOutOfProcessFetcher(PrivateMethods) helperDidTerminate:] The Internet connection appears to be offline. [NSURLErrorDomain:-1009] 12/27/13 10:49:43.243 PM GoogleSoftwareUpdateDaemon[53663]: -[KSServerUpdateRequest(PrivateMethods) fetcher:failedWithError:] KSServerUpdateRequest fetch failed. (productIDs: com.google.Keystone, ... (2)) [com.google.UpdateEngine.CoreErrorDomain:702 - 'htt[PeeEs]://tools.google.com/service/update2'] (The Internet connection appears to be offline. [NSURLErrorDomain:-1009]) 12/27/13 10:49:43.244 PM GoogleSoftwareUpdateDaemon[53663]: -[KSUpdateCheckAction(PrivateMethods) finishAction] KSUpdateCheckAction found updates: {( )} 12/27/13 10:49:43.247 PM GoogleSoftwareUpdateDaemon[53663]: -[KSPrefetchAction performAction] KSPrefetchAction no updates to prefetch. 12/27/13 10:49:43.248 PM GoogleSoftwareUpdateDaemon[53663]: -[KSMultiUpdateAction performAction] KSSilentUpdateAction had no updates to apply. 12/27/13 10:49:43.248 PM GoogleSoftwareUpdateDaemon[53663]: -[KSMultiUpdateAction performAction] KSPromptAction had no updates to apply. 12/27/13 10:49:43.250 PM GoogleSoftwareUpdateDaemon[53663]: -[KSUpdateEngine(PrivateMethods) updateFinish] KSUpdateEngine update processing complete. 12/27/13 10:49:45.570 PM GoogleSoftwareUpdateDaemon[53663]: -[KeystoneDaemon logServiceState] GoogleSoftwareUpdate daemon (1.1.0.3659) vending: com.google.Keystone.Daemon.UpdateEngine: 1 connection(s) com.google.Keystone.Daemon.Administration: 0 connection(s) 12/27/13 10:49:50.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: connect to the server /Volumes/Media Archive-1 12/27/13 10:49:50.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: connect on /Volumes/Media Archive-1 failed 65. 12/27/13 10:49:50.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: sleep for 10 seconds and then try again 12/27/13 10:49:53.828 PM KernelEventAgent[57]: tid 00000000 unmounting 1 filesystems 12/27/13 10:49:53.000 PM kernel[0]: AFP_VFS afpfs_unmount: /Volumes/Media Archive-1, flags 524288, pid 57 12/27/13 10:49:54.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: get the reconnect token 12/27/13 10:49:54.000 PM kernel[0]: AFP_VFS afpfs_DoReconnect: GetReconnectToken failed 32 /Volumes/Media Archive-1 12/27/13 10:49:54.000 PM kernel[0]: AFP_VFS afpfs_unmount : afpfs_DoReconnect sent signal for unmount to proceed 12/27/13 10:50:12.104 PM GoogleSoftwareUpdateDaemon[53663]: -[KeystoneDaemon main] GoogleSoftwareUpdateDaemon inactive, shutdown. 12/27/13 10:50:29.396 PM Dock[93157]: no information back from LS about running process

    Read the article

  • How does Windows 7 taskbar "color hot-tracking" feature calculate the colour to use?

    - by theyetiman
    This has intrigued me for quite some time. Does anyone know the algorithm Windows 7 Aero uses to determine the colour to use as the hot-tracking hover highlight on taskbar buttons for currently-running apps? It is definitely based on the icon of the app, but I can't see a specific pattern of where it's getting the colour value from. It doesn't seem to be any of the following: An average colour value from the entire icon, otherwise you would get brown all the time with multi-coloured icons like Chrome. The colour used the most in the image, otherwise you'd get yellow for the SQL Server Management Studio icon (6th from left). Also, the Chrome icon used red, green and yellow in equal measure. A colour located at certain pixel coordinates within the icon, because Chrome is red -indicating the top of the icon - and Notepad++ (2nd from right) is green - indicating the bottom of the icon. I asked this question on ux.stackoverflow.com and it got closed as off-topic, but someone answered with the following: As described by Raymond Chen in this MSDN blog article: Some people ask how it's done. It's really nothing special. The code just looks for the predominant color in the icon. (And, since visual designers are sticklers for this sort of thing, black, white, and shades of gray are not considered "colors" for the purpose of this calculation.) However I wasn't really satisfied with that answer because it doesn't explain how the "predominant" colour is calculated. Surely on the SQL Management Studio icon, the predominant colour, to my eyes at least, is yellow. Yet the highlight is green. I want to know, specifically, what the algorithm is.

    Read the article

  • BIND master/slave does not respond for queries for its slave

    - by Savas
    Systems are all Centos 6.2 Lets say I have a masterdns with IP 10.2.1.2, authoritative for the 10.2.1.X subnet and let say it is domain example.com I have another two subnets, 10.2.2.X and 10.1.2.X Each one has its own DNS server, dns2 and dns1 respectively and let say these are domains dom2.example.com and dom2.example.com respectively. The masterdns server has slave zones for dns1, dns2 and respond to requests OK. The dns1, dns2 have the masterdns zones as slaves two, and respond to requests OK. So, the masterdns has as slave zones all the subordinate domains of example.com Each of dns1 and dns2 use masterdns as a forwards (which uses another dns cache/proxy server) for dns resolution of internet public domain names. It works OK that too. The problem is, and I cannot figure it out. Why queries for example at dns1 for hostnames of dom2.example.com do not resolve? If i use nslookup - masterdns at dns1 server, resolve (i use directly the dns facility of masterdns). If I use nslookup locally, meaning queries are sent to dns1, for hosts that are at dom2.example.com, they do not resolve. Everything other works OK.

    Read the article

  • Squid: The request or reply is too large

    - by Ueli
    I have done a reverse proxy with an Apache in the background (on the same server). All works great but I can't open one page. I get the error "The request or reply is too large." In my cache.log contains: 2010/12/09 15:28:29| WARNING: http.c:971: HTTP header too large 2010/12/09 15:29:03| ctx: enter level 0: 'http://server/admin/cms/nav' 2010/12/09 15:29:03| httpProcessReplyHeader: Too large reply header 2010/12/09 15:29:03| ctx: exit level 0 In my squid.conf i disabled the limitations of the request and reply header, without success: reply_body_max_size 0 allow all request_body_max_size 0 Does someone know why that don't work? Thank you very much. Squid Version: Squid Cache: Version 2.7.STABLE3 configure options: '--prefix=/usr' '--exec_prefix=/usr' '--bindir=/usr/sbin' '--sbindir=/usr/sbin' '--libexecdir=/usr/lib/squid' '--sysconfdir=/etc/squid' '--localstatedir=/var/spool/squid' '--datadir=/usr/share/squid' '--enable-async-io' '--with-pthreads' '--enable-storeio=ufs,aufs,coss,diskd,null' '--enable-linux-netfilter' '--enable-arp-acl' '--enable-epoll' '--enable-removal-policies=lru,heap' '--enable-snmp' '--enable-delay-pools' '--enable-htcp' '--enable-cache-digests' '--enable-underscores' '--enable-referer-log' '--enable-useragent-log' '--enable-auth=basic,digest,ntlm,negotiate' '--enable-negotiate-auth-helpers=squid_kerb_auth' '--enable-carp' '--enable-follow-x-forwarded-for' '--with-large-files' '--with-maxfd=65536' 'amd64-debian-linux' 'build_alias=amd64-debian-linux' 'host_alias=amd64-debian-linux' 'target_alias=amd64-debian-linux' 'CFLAGS=-Wall -g -O2' 'LDFLAGS=' 'CPPFLAGS='

    Read the article

  • Juniper’s Network Connect ncsvc on Linux: “host checker failed, error 10”

    - by hfs
    I’m trying to log in to a Juniper VPN with Network Connect from a headless Linux client. I followed the instructions and used the script from http://mad-scientist.us/juniper.html. When running the script with --nogui switch the command that gets finally executed is $HOME/.juniper_networks/network_connect/ncsvc -h HOST -u USER -r REALM -f $HOME/.vpn.default.crt. I get asked for the password, a line “Connecting to…” is printed but then the programm silently stops. When adding -L 5 (most verbose logging) to the command line, these are the last messages printed to the log: dsclient.info state: kStateCacheCleaner (dsclient.cpp:280) dsclient.info --> POST /dana-na/cc/ccupdate.cgi (authenticate.cpp:162) http_connection.para Entering state_start_connection (http_connection.cpp:282) http_connection.para Entering state_continue_connection (http_connection.cpp:299) http_connection.para Entering state_ssl_connect (http_connection.cpp:468) dsssl.para SSL connect ssl=0x833e568/sd=4 connection using cipher RC4-MD5 (DSSSLSock.cpp:656) http_connection.para Returning DSHTTP_COMPLETE from state_ssl_connect (http_connection.cpp:476) DSHttp.debug state_reading_response_body - copying 0 buffered bytes (http_requester.cpp:800) DSHttp.debug state_reading_response_body - recv'd 0 bytes data (http_requester.cpp:833) dsclient.info <-- 200 (authenticate.cpp:194) dsclient.error state host checker failed, error 10 (dsclient.cpp:282) ncapp.error Failed to authenticate with IVE. Error 10 (ncsvc.cpp:197) dsncuiapi.para DsNcUiApi::~DsNcUiApi (dsncuiapi.cpp:72) What does host checker failed mean? How can I find out what it tried to check and what failed? The HostChecker Configuration Guide mentions that a $HOME/.juniper_networks/tncc.jar gets installed on Linux, but my installation contains no such file. From that I concluded that HostChecker is disabled for my VPN on Linux? Are the POST to /dana-na/cc/ccupdate.cgi and “host checker failed” connected or independent? By running the connection over a SSL proxy I found out that the POST data is status=NOTOK (Funny side note: the client of the oh-so-secure VPN does not validate the server’s SSL certificate, so is wide open to MITM attacks…). So it seems that it’s the client that closes the connection and not the server.

    Read the article

  • Anyone know a good mind mapper that works with a scheduler?

    - by GLycan
    TL;DR: Mind mapping tasks to be processed into a schedule based on task metadata. I have all sorts of ideas about what to invest resources (mainly time) in, but when I actually have time to do something I useually end up browsing reddit for not knowing what do to, and the frequancy with which I forget deadlines scares me. I'd love to bring order and structure into my mind, and always know what to do next. So, I want a mind mapping app, where I'd give each branch (types and subtypes of things I want to do) a importance score (if there were two branches, and one had 60 while the other 40, they would respectivily get 60% and 40% of the parent's importance, with the root being 100) and a how soon that branch should be revised/updated (an hobby I want to try out might be checked, say, once a week, while a school subject should be checked once a day) and give each leaf (something I want/need to do) how much time it takes, deadline (if any), and optionally an absolute importance, reoccurrence (guitar practice might repeat once a week), and prerequisites (reading something requires that book (although that could be brought somewhere), coding requires a box, jogging requires being outside) and maybe some other flags, like if it's enjoyable or not. It should either be packaged or working with a schedular app, to which I'd say, look, my day works this way (completely busy from 8 to 9:15, then 15 minutes of being inside with nothing, ..., two hours with box and possibility to go outside, etc), saying that such-and-such pattern is school and happens ever weekday except such-and-such days. The output should be of the form of a schedule, fit for printing or, when I finally get an android, mobile viewing, that schedules tasks with regards to availability of resources and importance (importance being derived from the leaf-task's parent branches), and the set of flags (all work and no play makes me a dull boy). One of these tasks should be reviewing anything that should be updated on that day, including future day layouts (e.g, if the time slots of future days have changed. This should be done every day.) Does anyone know some collection of preferably open-source (or free, or pirateable) tools, or better yet a single one, that accomplishes this task? I know python pretty well, and should be able to write any necessary glue.

    Read the article

  • Is my webserver being abused for banking fraud?

    - by koffie
    Since a few weeks i'm getting a lot of 403 errors from apache in my log files that seem to be related to a bank frauding scheme. The relevant log entries look like this (The ip 1.2.3.4 is one I made up, I did not modify the rest of each line) www.bradesco.com.br:80 / 1.2.3.4 - - [01/Dec/2012:07:20:32 +0100] "GET / HTTP/1.1" 403 427 "-" "Mozilla/5.0 (Windows NT 5.1) AppleWebKit/535.11 (KHTML, like Gecko) Chrome/17.0.963.56 Safari/535.11" www.bb.com.br:80 / 1.2.3.4 - - [01/Dec/2012:07:20:32 +0100] "GET / HTTP/1.1" 403 370 "-" "Mozilla/5.0 (Windows NT 5.1) AppleWebKit/535.11 (KHTML, like Gecko) Chrome/17.0.963.56 Safari/535.11" www.santander.com.br:80 / 1.2.3.4 - - [01/Dec/2012:07:20:33 +0100] "GET / HTTP/1.1" 403 370 "-" "Mozilla/5.0 (Windows NT 5.1) AppleWebKit/535.11 (KHTML, like Gecko) Chrome/17.0.963.56 Safari/535.11" www.banese.com.br:80 / 1.2.3.4 - - [01/Dec/2012:07:20:33 +0100] "GET / HTTP/1.1" 403 370 "-" "Mozilla/5.0 (Windows NT 5.1) AppleWebKit/535.11 (KHTML, like Gecko) Chrome/17.0.963.56 Safari/535.11" the logformat I use is: LogFormat "%V:%p %U %h %l %u %t \"%r\" %>s %O \"%{Referer}i\" \"%{User-Agent}i\"" The strange thing is that all these domains are domains of banks and 3 out of the 4 domains are also in the list of the bank frauding scheme described on: http://www.abuse.ch/?p=2925 I would really like to know if my server is being abused for bank frauding or not. I suspect not, because it's giving 403 to all requests. But any extra checks that I can do to ensure that my server is not being abused are welcome. I'm also curious on how the "bad guys" expected my server to behave. I.e. are they just expecting my server to act as a proxy to hide the ip of the fake site, or are they expecting that my server will actually serve the fake banking website? Is the ip 1.2.3.4 more likely to be the ip of a victim or the ip of a bad guy. I suspect a bad guy, because it's quite unlikely that a real person would visit 4 bank sites in a second. If it's from a bad guy I'm very curious at what he is trying to do.

    Read the article

  • AWS VPC ELB vs. Custom Load Balancing

    - by CP510
    So I'm wondering if this is a good idea. I have a Amazon AWS VPC setup with a public and private subnets. So I all ready get the Internet Gateway and NAT. I was going to setup all my web servers (Apache2 isntances) and DB servers in the private subnet and use a Load Balancer/Reverse Proxy to pick up requests and send them into the private subnets cluster of servers. My question then, is Amazons ELB's a good use for these, or is it better to setup my own custom instance to handle the public requests and run them through the NAT using nginx or pound? I like the second option just for the sake of having a instance I can log into and check. As well as taking advantage of caching and fail2ban ddos prevention, as well as possibly using fail safes to redirect traffic. But I have no experience with their ELB's, so I thought I'd ask your opinions. Also, if you guys have an opinion on this as well, would using the second option allow me to only have 1 public IP address and be able to route SSH connections through port numbers to respective instances? Thanks in advance!

    Read the article

  • Howto disable SSH local port forwarding ?

    - by SCO
    I have a server running Ubuntu and the OpenSSH daemon. Let's call it S1. I use this server from client machines (let's call one of them C1) to do an SSH reverse tunnel by using remote port forwarding, eg : ssh -R 1234:localhost:23 login@S1 On S1, I use the default sshd_config file. From what I can see, anyone having the right credentials {login,pwd} on S1 can log into S1 and either do remote port forwarding and local port forwarding. Such credentials could be a certificate in the future, so in my understanding anyone grabbing the certificate can log into S1 from anywhere else (not necessarily C1) and hence create local port forwardings. To me, allowing local port forwarding is too dangerous, since it allows to create some kind of public proxy. I'm looking for a way tto disable only -L forwardings. I tried the following, but this disables both local and remote forwarding : AllowTcpForwarding No I also tried the following, this will only allow -L to SX:1. It's better than nothing, but still not what I need, which is a "none" option. PermitOpen SX:1 So I'm wondering if there is a way, so that I can forbid all local port forwards to write something like : PermitOpen none:none Is the following a nice idea ? PermitOpen localhost:1

    Read the article

  • IIS 7 rewriting subdomain to point at a specific port.

    - by Tommy Jakobsen
    Having installed Team Foundation Server 2010 on Windows Server 2008, I need an easy URL for our developers to access their repositories. The default URL for the TFS repositories is http://localhost:8080/tfs Now I want the subdomain domain tfs.server.domain.com to point at http://localhost:8080/tfs. And when you write access tfs.server.domain.com/repos_name it should redirect to http://localhost:8080/tfs/repos_name. How can I do this in IIS 7? I already tried using the following rule, but it does not work. I get a 404. <rewrite> <globalRules> <rule name="TFS" stopProcessing="true"> <match url="^(?:tfs/)(.*)" /> <conditions> <add input="{HTTP_HOST}" pattern="^tfs.server.domain.com$" /> </conditions> <action type="Rewrite" url="http://localhost:8080/tfs/{R:1}" /> </rule> </globalRules> </rewrite>

    Read the article

  • .NET 2.0 Application now running slow on IIS 7.5

    - by Valien
    I recently moved (and still in testing) an application from a Windows 2003 Server (Physical box) running IIS 6.x to a Windows 2008 R2 Standard (VM) IIS 7.5 server. The application is a .NET framework 2.0 application and is running under a 2.0 App Pool. This site works great except for one thing: Takes forever to get a request back. I've been tracking it with Chrome Inspect Element and it queries the site and can take up to 45 seconds to answer. Now when it does the page(s) render instantly but it's that initial request that's killing it. I see no error logs or issues with the application or Windows Event Viewer or even IIS logs so not sure where to start looking next. Some new changes was that previously the app resided behind a Pix firewall and now is behind a larger network environment in a DMZ zone (and I believe NetScaler is also being used to manage the network). I do not have rights/abilities to look at the network itself but can contact the Data center folks to look deeper into this but I wanted to make sure it's not my application that might be causing the slowdown or IIS. In summary: .NET 2.0 application works great in IIS 6.x Application moved to an IIS 7.5 server and now slow on rendering but when it does render responds back with pages instantly. Edit for solution Found out that it was the SOAP calls that were slowing the site down. In the new datacenter my application cannot request SOAP calls and so they time out after 40-45 seconds or so. Now trying to find out if I can install a proxy server to redirect this...

    Read the article

  • SMTP message rate control on Ubuntu 8.04, preferably with postfix

    - by TimDaMan
    Maybe I am chasing a bug but I am trying to set up a smtp proxy of sorts. I have a postfix server which receives all the email for a collection of servers/clients. It them uses a smarthost (relayhost=...) to forward it's mail to our corporate MTA. I would like to limit the number of messages an individual server can relay to prevent swamping the corporate MTA. Postfix has a program called "anvil" that is capable of tracking stats about mail to be used for such things but it doesn't seem to be executed. I ran "inotifywait -m /usr/lib/postfix/anvil" while I started postfix and sent a number of messages through it from a remote server. inotifywait indicated anvil was never run. Anyone gotten postfix/anvil rate controls to work? main.cf smtpd_banner = $myhostname ESMTP $mail_name (Ubuntu) biff = no append_dot_mydomain = no readme_directory = no myhostname = site-server-q9 alias_maps = hash:/etc/aliases alias_database = hash:/etc/aliases myorigin = /etc/mailname mydestination = localhost relayhost = Out outgoing mail relay mynetworks = 127.0.0.0/8 [::ffff:127.0.0.0]/104 [::1]/128 10.0.0.0/8 mailbox_size_limit = 0 recipient_delimiter = + inet_interfaces = 10.X.X.X smtpd_client_message_rate_limit = 1 anvil_rate_time_unit = 1h master.cf extract anvil unix - - - - 1 anvil smtp inet n - - - - smtpd

    Read the article

  • memory tuning with rails/unicorn running on ubuntu

    - by user970193
    I am running unicorn on Ubuntu 11, Rails 3.0, and Ruby 1.8.7. It is an 8 core ec2 box, and I am running 15 workers. CPU never seems to get pinned, and I seem to be handling requests pretty nicely. My question concerns memory usage, and what concerns I should have with what I am seeing. (if any) Here is the scenario: Under constant load (about 15 reqs/sec coming in from nginx), over the course of an hour, each server in the 3 server cluster loses about 100MB / hour. This is a linear slope for about 6 hours, then it appears to level out, but still maybe appear to lose about 10MB/hour. If I drop my page caches using the linux command echo 1 /proc/sys/vm/drop_caches, the available free memory shoots back up to what it was when I started the unicorns, and the memory loss pattern begins again over the hours. Before: total used free shared buffers cached Mem: 7130244 5005376 2124868 0 113628 422856 -/+ buffers/cache: 4468892 2661352 Swap: 33554428 0 33554428 After: total used free shared buffers cached Mem: 7130244 4467144 2663100 0 228 11172 -/+ buffers/cache: 4455744 2674500 Swap: 33554428 0 33554428 My Ruby code does use memoizations and I'm assuming Ruby/Rails/Unicorn is keeping its own caches... what I'm wondering is should I be worried about this behaviour? FWIW, my Unicorn config: worker_processes 15 listen "#{CAPISTRANO_ROOT}/shared/pids/unicorn_socket", :backlog = 1024 listen 8080, :tcp_nopush = true timeout 180 pid "#{CAPISTRANO_ROOT}/shared/pids/unicorn.pid" GC.respond_to?(:copy_on_write_friendly=) and GC.copy_on_write_friendly = true before_fork do |server, worker| STDERR.puts "XXXXXXXXXXXXXXXXXXX BEFORE FORK" print_gemfile_location defined?(ActiveRecord::Base) and ActiveRecord::Base.connection.disconnect! defined?(Resque) and Resque.redis.client.disconnect old_pid = "#{CAPISTRANO_ROOT}/shared/pids/unicorn.pid.oldbin" if File.exists?(old_pid) && server.pid != old_pid begin Process.kill("QUIT", File.read(old_pid).to_i) rescue Errno::ENOENT, Errno::ESRCH # already killed end end File.open("#{CAPISTRANO_ROOT}/shared/pids/unicorn.pid.ok", "w"){|f| f.print($$.to_s)} end after_fork do |server, worker| defined?(ActiveRecord::Base) and ActiveRecord::Base.establish_connection defined?(Resque) and Resque.redis.client.connect end Is there a need to experiment enforcing more stringent garbage collection using OobGC (http://unicorn.bogomips.org/Unicorn/OobGC.html)? Or is this just normal behaviour, and when/as the system needs more memory, it will empty the caches by itself, without me manually running that cache command? Basically, is this normal, expected behaviour? tia

    Read the article

  • White Screen, No Errors.

    - by GruffTech
    So.. Interesting problem for you guys, As I'm completely lost as to what to do, or where to take the next step. Server & Application Environment. CentOS release 5.3 (Final) Apache 2.2.3-22 EnableSendfile off EnableMMAP off ErrorLog logs/error_log LogLevel debug PHP-5.2.6-2 error_reporting = E_ALL display_errors = on log_errors = on max_execution_time=300 max_input_time=60 memory_limit=512mb Kohana 2.3 PHP Environment. HAProxy 1.3.15.6-2 MemCacheD 1.2.6-1 Our application is split between 3 web servers, mounting a NFS Storage server, and sticky load balancing between the 3 web servers. The application seemingly runs great, but every so often, instead of loading, the application just shows a pure white page. Not a 404 Error, or a 500 Server Error, a clean white page. And it returns instantly, so its not a execution time error. Nothing in the Error log, or Server-Error Log, Proxy log shows standard proxied connection, Just the standard 200-Status in Access log, with 256 bytes transferred. To me, this leads to tell me that the application itself is having a problem. A rare, unexplainable, seemingly random, problem that causes what we've now called the "White Screen of Death." Our developers all say that since there is nothing going to our error logs, that it must be a server problem. But I say the same thing, There's nothing going to ANY of our logs (relevent to this anyway), and we're not having httpd children crash from what i can tell. Any ideas on how i can increase my logs, or somehow prove that its not a bug in PHP, Apache, CentOS, ect? Or if it is somehow a bug, identify it?

    Read the article

  • chrooted sftp user with write permissions to /var/www

    - by matthew
    I am getting confused about this setup that I am trying to deploy. I hope someone of you folks can lend me a hand: much much appreciated. Background info Server is Debian 6.0, ext3, with Apache2/SSL and Nginx at the front as reverse proxy. I need to provide sftp access to the Apache root directory (/var/www), making sure that the sftp user is chrooted to that path with RWX permissions. All this without modifying any default permission in /var/www. drwxr-xr-x 9 root root 4096 Nov 4 22:46 www Inside /var/www -rw-r----- 1 www-data www-data 177 Mar 11 2012 file1 drwxr-x--- 6 www-data www-data 4096 Sep 10 2012 dir1 drwxr-xr-x 7 www-data www-data 4096 Sep 28 2012 dir2 -rw------- 1 root root 19 Apr 6 2012 file2 -rw------- 1 root root 3548528 Sep 28 2012 file3 drwxr-x--- 6 www-data www-data 4096 Aug 22 00:11 dir3 drwxr-x--- 5 www-data www-data 4096 Jul 15 2012 dir4 drwxr-x--- 2 www-data www-data 536576 Nov 24 2012 dir5 drwxr-x--- 2 www-data www-data 4096 Nov 5 00:00 dir6 drwxr-x--- 2 www-data www-data 4096 Nov 4 13:24 dir7 What I have tried created a new group secureftp created a new sftp user, joined to secureftp and www-data groups also with nologin shell. Homedir is / edited sshd_config with Subsystem sftp internal-sftp AllowTcpForwarding no Match Group <secureftp> ChrootDirectory /var/www ForceCommand internal-sftp I can login with the sftp user, list files but no write action is allowed. Sftp user is in the www-data group but permissions in /var/www are read/read+x for the group bit so... It doesn't work. I've also tried with ACL, but as I apply ACL RWX permissions for the sftp user to /var/www (dirs and files recursively), it will change the unix permissions as well which is what I don't want. What can I do here? I was thinking I could enable the user www-data to login as sftp, so that it'll be able to modify files/dirs that www-data owns in /var/www. But for some reason I think this would be a stupid move securitywise.

    Read the article

  • Authority Information Access local path being ignored

    - by Kevin
    I have a CA set up in Server 2008 R2, and generally it is working, but I can't control the local path/filename it writes its own certificate to for the Authority Information Access publishing. Here's a screen shot of the dialog I'm trying to set this on: From these settings I would expect to get the file: C:\Windows\system32\CertSrv\CertEnroll\DAMNIT.crt But instead I get: C:\Windows\system32\CertSrv\CertEnroll\SERVER.domain.com_My Issuing Authority(1).crt Of course, the actual change shown wouldn't be very useful, but it's illustrative; no matter what path/filename I use, it always lands up in the same place and with the same name. I actually wanted to change the name from <ServerDNSName>_<CaName><CertificateName>.crt to <CaName><CertificateName>.crt, since the latter corresponds to the HTTP URL whereas the former does not. Admittedly, I haven't set up many CAs so perhaps I'm just deluded as to what this dialog is supposed to be setting, but if so this is notoriously bad UI design. (Incidentally, I have a couple other complaints with the same dialog.) What's going on here and is there some way to get the filename pattern I want?

    Read the article

  • VMWare steals IP addresses

    - by Ishan Amin
    I'm having a peculiar problem, that I think I have narrowed down to VMware. For the past one year, every once in a while we lose internet connection and not all users (about 10 users) go down at the same time, its usually one-by-one. First someone will call me and say "Internet is down" and then we would go reset the router and modem and switch and it would be working again for a while, then go down again without any pattern or replicatable sequence. We'd go repeat the steps again to get everyone in the office running again. We called our Internet Service Provider and they constantly say, We see your modem and we see your router and from thier end everything is OK. we replaced our router and switch and modem, twice! Last friday, it dawned upon me, that everytime we turn on a VMware machine, this sequence of taking everyone down starts, which also explains the message that my users get for "IP Conflict Found" So we do alot of VMware testing and lo and behold, it takes my Internet down. My Yahoo and Gtalk would continue working but www is down when the VMware machines are started. I do use bridged networking to all the VMware machines, but I dont know what else to set it at. now, sorry for this long rambling but anyone have any clue on how to stop this? thanks IA

    Read the article

  • Windows Network File Transfer to Samba server: “Are you sure you want to copy this file without its properties?”

    - by jimp
    I am transferring a lot of files to a new NAS based on OpenMediaVault, with the Samba 3.5.6 service running. I am transferring from Windows 7 64-bit to the NAS, and on some media files Windows is prompting about losing some property data across the transfer. I have never seen this before when transferring to Samba boxes I have built myself (vs this turnkey solution), so I'm guessing there must be a Samba setting I can change to preserve the file properties in question instead of permanently losing whatever they contain (Date Taken? Exposure? Flash Fired? etc). Or maybe I've just never encountered this before; I'm really not sure. I tried adding ea support = yes and store dos attributes = yes to the [global] section, but the problem remains. The Linux file system is ext4 mounted with user_xattr (full options: defaults,acl,user_xattr,noexec,usrjquota=aquota.user,grpjquota=aquota.group,jqfmt=vfsv0) as Samba requires. Any ideas would be greatly appreciated. Thank you! Samba config: [global] workgroup = WORKGROUP server string = %h server include = /etc/samba/dhcp.conf dns proxy = no log level = 2 syslog = 2 log file = /var/log/samba/log.%m max log size = 1000 syslog only = yes panic action = /usr/share/samba/panic-action %d encrypt passwords = true passdb backend = tdbsam obey pam restrictions = yes unix password sync = no passwd program = /usr/bin/passwd %u passwd chat = *Enter\snew\s*\spassword:* %n\n *Retype\snew\s*\spassword:* %n\n *password\supdated\ssuccessfully* . pam password change = yes socket options = TCP_NODELAY IPTOS_LOWDELAY guest account = nobody load printers = no disable spoolss = yes printing = bsd printcap name = /dev/null unix extensions = yes wide links = no create mask = 0777 directory mask = 0777 use sendfile = no null passwords = no local master = yes time server = yes wins support = yes ea support = yes store dos attributes = yes Note: I found this related question, but it explains the loss due to the user trying to transfer from NTFS to FAT32.

    Read the article

  • How to set up simple VPN for secure Internet connections over unencrypted Wi-Fi on Windows?

    - by Senseful
    I'm looking for a solution similar to the one in this question, except that I don't have a linux computer. I have windows computers that could be set up to accept VPN connections. Preferably I want to set this up on either Windows Server 2003 or Windows XP. I'd like to connect different devices (e.g. iPhone, iPad, laptops, etc.) that are on open unsecure wireless networks (e.g. the one's you see at places like Starbucks) to this VPN to ensure that all my data is secure. I found an article that shows that you can enable VPN connections on Windows XP. After following those steps, though, I'm not sure what to do. Which ports do I open on my firewall? Which VPN settings do I use on my devices such as the iPhone? Do I use L2TP, PPTP, or IPSec? What's the difference between these? Are there any other steps missing in that tutorial? I'm hoping that since Windows has this built in feature, that it will be much simpler to set up rather than having to deal with setting up something such as OpenVPN. If I follow those settings and enable port forwarding on port 1723, and then use the following settings on the iPhone: PPTP (IP Address) RSA SecurID: Off Encryption Level: Auto Send All Traffic: On Proxy: Off It shows "Connecting..." then "Disconnecting..." and the following error message: VPN Configuration A connection could not be established to the PPP server. Try reconnecting. If the problem continues, verify your settings and contact your Administrator. I'm using a user account that I enabled privileges to in the VPN settings on the Windows machine.

    Read the article

  • IE6 does not follow 302 redirect - displays 404 instead

    - by Dexter
    One of our clients has reported that they are experiencing 404 (file not found) errors when attempting to navigate a website that we support. The behaviour only appears to affect her - other users on the same machine can navigate the website fine, but the problem follows her from one PC to another. I've had a good look through the IIS server logs and have identified the requests in question. The normal request pattern is as follows: POST /page.aspx - 80 - ... 401 1 0 POST /page.aspx - 80 DOMAIN/user ... 302 0 0 GET /anotherPage.aspx Request=833f80a5-f34c-4b0e-addb-d73e1ee1663a 80 - ... 401 1 0 GET /anotherPage.aspx Request=833f80a5-f34c-4b0e-addb-d73e1ee1663a 80 DOMAIN/user ... 200 0 However, requests for the affected user do not include a request for the redirected page, nor an entry for the 404, i.e.: POST /page.aspx - 80 - ... 401 1 0 POST /page.aspx - 80 DOMAIN/user ... 302 0 0 ... other unrelated requests Can anyone suggest what might trigger this behaviour, and how I might investigate the cause or prevent it from occuring? I read here that the Allow META refresh option in IE6 might trigger this behaviour, but I have not been able to replicate the behaviour by modifying this setting only.

    Read the article

  • nginx: dump HTTP requests for debugging

    - by Alexander Gladysh
    Ubuntu 10.04.2 nginx 0.7.65 I see some weird HTTP requests coming to my nginx server. To better understand what is going on, I want to dump whole HTTP request data for such queries. (I.e. dump all request headers and body somewhere I can read them.) Can I do this with nginx? Alternatively, is there some HTTP server that allows me to do this out of the box, to which I can proxy these requests by the means of nginx? Update: Note that this box has a bunch of normal traffic, and I would like to avoid capturing all of it on low level (say, with tcpdump) and filtering it out later. I think it would be much easier to filter good traffic first in a rewrite rule (fortunately I can write one quite easily in this case), and then deal with bogus traffic only. And I do not want to channel bogus traffic to another box just to be able to capture it there with tcpdump. Update 2: To give a bit more details, bogus request have parameter named (say) foo in their GET query (the value of the parameter can differ). Good requests are guaranteed not to have this parameter ever. If I can filter by this in tcpdump or ngrep somehow — no problem, I'll use these.

    Read the article

  • How should I configure nginx caching headers for a "baked" static file blog? (Octopress)

    - by Doug Stephen
    I recently deployed an Octopress blog (which is a blogging platform built around Jekyll). It's a static-site blog generator, with no dynamic content or databases to muck about with. It's being served up by nginx. My question is, what is the appropriate expires directive or Cache-Control header that I should set to make sure that visitors get the most up-to-date version of the site when they visit without having to manually refresh? Since the site is just .html files it seems to get cached pretty aggressively. I've tried a million different combinations of expires modified + xxxx and even straight up expires off but I can't seem to wrap my head around it. I'm very new to dealing with caching like this, specifically, on static files that change frequently, and obviously if the site hasn't been changed then I'd like for it to be served up out of the cache. Update (still not solved though): I found open_file_cache, tweaked that. Still no dice. It seems like what I might want to do is use nginx as a proxy cache and use Apache with ETags? Is there really no convenient way to make nginx play nicer with conditional requests from the client? TL;DR: I'm running a static-file blog and I'd like to set up nginx to only serve from the cache if the blog hasn't been updated recently, but I'm too stupid to figure it out myself because I'm relatively new to web servers.

    Read the article

  • Intermittent 5.7.1 email bounce to Exchange 2007

    - by Steve Kennaird
    My knowledge of Exchange isn't particularly great, so excuse me if some of the terminology I use isn't quite right. I'm primarily a web developer who's now responsible for a small business's network. We have a server running SBS 2008 and Exchange 2007. Generally, everything works well, emails are able to be sent to both internal and external domains without issue. We've only got ~20 users, Exchange is sitting on a single server. I use SendGrid to send emails generated by our externally hosted website to users in the office. Primarily, order notifications are sent to [email protected]. Without any pattern and less than once per week on average, an email to [email protected] will bounce back, and the logs on SendGrid detail the following error: 550 5.7.1 Unable to relay for [email protected] Either side of that failed delivery attempt, I'm able to send and receive emails to/from [email protected]. Having done some research, incorrect reverse DNS seems like it could be a cause of intermittent bounces like this. Having used nslookup, I have found that the reverse DNS doesn't map like it should, e.g. Office IP: 135.325.351.123 (made up IP, for example only) Domain: office.somedomain.com (made up, for example only) Reverse DNS: somedomain.gotadsl.co.uk (half made up) Could this be a cause? I'm sure that the IP address and the domain should map to each other. Also, it has been suggested to me that as the Exchange server is on a network with an ADSL connection, that could be a potential cause as the connection "goes up and down all day long". I don't have an opinion on this, as I don't have enough knowledge of Exchange/ADSL to form a reliable opinion. Can anyone offer any insight as to whether one or both are actually potential causes, or if there is another possible cause?

    Read the article

  • How to find out what is causing a slow down of the application on this server?

    - by Jan P.
    This is not the typical serverfault question, but I'm out of ideas and don't know where else to go. If there are better places to ask this, just point me there in the comments. Thanks. Situation We have this web application that uses Zend Framework, so runs in PHP on an Apache web server. We use MySQL for data storage and memcached for object caching. The application has a very unique usage and load pattern. It is a mobile web application where every full hour a cronjob looks through the database for users that have some information waiting or action to do and sends this information to a (external) notification server, that pushes these notifications to them. After the users get these notifications, the go to the app and use it, mostly for a very short time. An hour later, same thing happens. Problem In the last few weeks usage of the application really started to grow. In the last few days we encountered very high load and doubling of application response times during and after the sending of these notifications (so basically every hour). The server doesn't crash or stop responding to requests, it just gets slower and slower and often takes 20 minutes to recover - until the same thing starts again at the full hour. We have extensive monitoring in place (New Relic, collectd) but I can't figure out what's wrong; I can't find the bottlekneck. That's where you come in: Can you help me figure out what's wrong and maybe how to fix it? Additional information The server is a 16 core Intel Xeon (8 cores with hyperthreading, I think) and 12GB RAM running Ubuntu 10.04 (Linux 3.2.4-20120307 x86_64). Apache is 2.2.x and PHP is Version 5.3.2-1ubuntu4.11. If any configuration information would help analyze the problem, just comment and I will add it. Graphs info phpinfo() apc status memcache status collectd Processes CPU Apache Load MySQL Vmem Disk New Relic Application performance Server overview Processes Network Disks (Sorry the graphs are gifs and not the same time period, but I think the most important info is in there)

    Read the article

< Previous Page | 312 313 314 315 316 317 318 319 320 321 322 323  | Next Page >