Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
At Tuesday 29/8/2006 10:45, bwhicks at aep.com wrote:
>We have cvsnt 2.5.02.2099 running on redhat ES3.7. Almost all of our
>clients are running windows and we use the sspi protocol with winbind.
>
>One of our groups uses Cruise Control for automatic building. The userid
>used for this process is local to the linux box and uses sserver for
>connecting. When a build kicks off, the load on the linux box goes up to
>around 1.6 and other users' sspi authentication requests fail.
>
>Does anybody have any ideas about the cause or a possible workaround? A
>single user shouldn't be able to lock everyone else out like that...
This old message may be of interest:
>Subject: RE: [cvsnt] Strange CVS performance fluctuation
>Date: Wed, 21 Jun 2006 08:58:57 +1200
>Message-ID: <FB1D5769F12AF54EB58B29B12517B2650193AA9E at svmxc301.ap.aderant.com>
>X-MS-Has-Attach:
>X-MS-TNEF-Correlator:
>Thread-Topic: [cvsnt] Strange CVS performance fluctuation
>From: "Trevor Leybourne" <trevor.leybourne at aderant.com>
>To: "Tobias Rodenbach" <tr.mailinglists at gmail.com>, <cvsnt at cvsnt.org cvsnt downloads at march-hare.com @CVSNT on Twitter CVSNT on Facebook>
>X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at nodomain.org
>X-Spam-Status: No, hits=-2.4 tagged_above=-1000.0 required=3.5 tests=AWL,
> BAYES_00, FORGED_RCVD_HELO
>X-Spam-Level:
>X-BeenThere: cvsnt at cvsnt.org cvsnt downloads at march-hare.com @CVSNT on Twitter CVSNT on Facebook
>X-Mailman-Version: 2.1.5
>Precedence: list
>List-Id: cvsnt Mailing List <cvsnt.cvsnt.org>
>List-Unsubscribe: <http://www.cvsnt.org/cgi-bin/mailman/listinfo/cvsnt https://www.march-hare.com/cvspro/en.asp#downcvs>,
> <mailto:cvsnt-request at cvsnt.org?subject=unsubscribe>
>List-Archive: <http://list.nodomain.org/pipermail/cvsnt>
>List-Post: <mailto:cvsnt at cvsnt.org cvsnt downloads at march-hare.com @CVSNT on Twitter CVSNT on Facebook>
>List-Help: <mailto:cvsnt-request at cvsnt.org?subject=help>
>List-Subscribe: <http://www.cvsnt.org/cgi-bin/mailman/listinfo/cvsnt https://www.march-hare.com/cvspro/en.asp#downcvs>,
> <mailto:cvsnt-request at cvsnt.org?subject=subscribe>
>
>
>Just some ideas and feedback if it will help,
>
>We've been having similar and significant problems here as well. Our CPU
>rarely goes above about 5-10% usage (even with 10-20 cvs.exe processes
>---- it's a fairly grunty Windows 2003 Server) and yet a checkout that
>would normally take less than 15 minutes will take more than a hour when
>multiple people are accessing it.
>
>Our IT guys have spent a lot of time evaluating the issue and the issues
>has been resolved as being disk contention (see timer results below).
>
>We found that running on a server with Raid 1 was costly and have moved
>the disk array to Raid 5 instead (no significant difference between SAN
>and LOCAL disk).
>
>Also, if you are using something like CruiseControl for your build
>process, look at the URL Build Trigger rather than the standared build
>trigger. The standard build trigger uses a CVS LOG command to look for
>changes and this is a significant process to run against a large
>repository.
>
>We ran a test of having 3 PCs each running 3 concurrent checkouts
>against CVS. Here are the results against different disk configurations
>and are in Minutes. The repository size is around 850Mb and a checked
>out Sandbox is 540Mb (6,000 files in 250 folders)
>
> (RAID5 Local) (RAID1 Local) (RAID5 SAN)
>Test1 14 77 15
>Test2 15 77 15
>Test3 14 76 15
>Test4 16 78 16
>Test5 16 77 16
>Test6 16 77 16
>Test7 16 72 17
>Test8 16 78 17
>Test9 16 79 16
>
>The CVS Server is 2.5.02.2115 and the client was 2.5.03.22650
>
>Cheers
>Trevor
>
>-----Original Message-----
>From: cvsnt-bounces at cvsnt.org [mailto:cvsnt-bounces at cvsnt.org] On Behalf
>Of Tobias Rodenbach
>Sent: Wednesday, 21 June 2006 1:41 a.m.
>To: cvsnt at cvsnt.org cvsnt downloads at march-hare.com @CVSNT on Twitter CVSNT on Facebook
>Subject: [cvsnt] Strange CVS performance fluctuation
>
>Hello,
>
>I observed the following strange behaviour, when "many" people tried to
>access our cvs server (I counted about 35 instances of cvs.exe).
>Then, the CPU usage history of the Windows Task Manager shows that CPU
>Usage goes to about 50% for about 10 seconds, then drops to below 3% for
>about 10 seconds, rises to about 50% for 10 seconds and so on...
>Sometimes CPU usage stays high or low for more than 10 sec but most of
>the time it alternates in the 10 second scheme.
>
>In general, the performance is very low. Checking out using a tag (which
>is the most common case here) can take up to one hour for around 6 MB in
>the local LAN. ("Normally" that takes significantly less than one
>minute.)
>
>The host is a Win2003 server with 2 hyperthreading CPUs, the cvsnt
>server is version 2.5.02 build 2115.
>
>Does anyone have any idea what could be the reason for this behaviour?
>
>Best regards
> Tobias
Gabriel Genellina
Softlab SRL
__________________________________________________
Preguntá. Respondé. Descubrí.
Todo lo que querías saber, y lo que ni imaginabas,
está en Yahoo! Respuestas (Beta).
¡Probalo ya!
http://www.yahoo.com.ar/respuestas