[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: [Subclipse-users] Slowness in svn+ssh

From: Bradley Wagner <bradley.wagner_at_hannonhill.com>
Date: 2006-05-04 22:48:30 CEST

Alexander,
The server machine is running SVN 1.3.1 and the OS X client machine
is using 1.3.1 for its binaries as well when running JavaHL. I'll
look for the sshd version, but because the problem is not limited to
svn+ssh:// alone, I don't think that's as essential. The OS X client
machine has no firewall or anti-virus software installed and is on
the same LAN as the server machine, and I've also tried running the
SVN server on the OS X machine itself and connecting via "localhost"
with the same results.

I don't think I can attach the log file as it is some 100 MB.

Do you think we can make the logging a little more restrictive so as
only to log JavaSVN debug logging?

The one thing I can see about the log file is the frequent occurrence
of the following exception:

May 4, 2006 4:15:51 PM
org.tmatesoft.svn.core.internal.util.DefaultSVNDebugLogger logInfo
FINE: svn: Malformed network data, string length expected
May 4, 2006 4:15:51 PM
org.tmatesoft.svn.core.internal.util.DefaultSVNDebugLogger logInfo
FINE: svn: Malformed network data, string length expected
java.lang.Exception: svn: Malformed network data, string length expected
        at org.tmatesoft.svn.core.internal.wc.SVNErrorManager.error
(SVNErrorManager.java:43)
        at org.tmatesoft.svn.core.internal.io.svn.SVNReader.readStringLength
(SVNReader.java:517)
        at org.tmatesoft.svn.core.internal.io.svn.SVNReader.readString
(SVNReader.java:501)
        at org.tmatesoft.svn.core.internal.io.svn.SVNReader.parse
(SVNReader.java:210)
        at
org.tmatesoft.svn.core.internal.io.svn.SVNEditModeReader.processCommand(
SVNEditModeReader.java:75)
        at org.tmatesoft.svn.core.internal.io.svn.SVNReader.parse
(SVNReader.java:306)
        at org.tmatesoft.svn.core.internal.io.svn.SVNConnection.read
(SVNConnection.java:199)
        at org.tmatesoft.svn.core.internal.io.svn.SVNRepositoryImpl.read
(SVNRepositoryImpl.java:887)
        at org.tmatesoft.svn.core.internal.io.svn.SVNRepositoryImpl.update
(SVNRepositoryImpl.java:493)
        at org.tmatesoft.svn.core.wc.SVNUpdateClient.doUpdate
(SVNUpdateClient.java:151)
        at org.tmatesoft.svn.core.wc.SVNUpdateClient.doCheckout
(SVNUpdateClient.java:283)
        at org.tmatesoft.svn.core.javahl.SVNClientImpl.checkout
(SVNClientImpl.java:293)
        at org.tmatesoft.svn.core.javahl.SVNClientImpl.checkout
(SVNClientImpl.java:302)
        at
org.tigris.subversion.svnclientadapter.javahl.AbstractJhlClientAdapter.c
heckout(AbstractJhlClientAdapter.java:213)
        at
org.tigris.subversion.subclipse.core.commands.CheckoutCommand.checkoutPr
oject(CheckoutCommand.java:190)
        at
org.tigris.subversion.subclipse.core.commands.CheckoutCommand.basicRun
(CheckoutCommand.java:157)
        at org.tigris.subversion.subclipse.core.commands.CheckoutCommand
$1.run(CheckoutCommand.java:210)
        at org.tigris.subversion.subclipse.core.SVNProviderPlugin$6.run
(SVNProviderPlugin.java:441)
        at org.eclipse.core.internal.resources.Workspace.run(Workspace.java:
1721)
        at org.tigris.subversion.subclipse.core.SVNProviderPlugin.run
(SVNProviderPlugin.java:436)
        at org.tigris.subversion.subclipse.core.commands.CheckoutCommand.run
(CheckoutCommand.java:208)
        at
org.tigris.subversion.subclipse.ui.operations.CheckoutAsProjectOperation
.execute(CheckoutAsProjectOperation.java:62)
        at
org.tigris.subversion.subclipse.ui.operations.CheckoutAsProjectOperation
.execute(CheckoutAsProjectOperation.java:47)
        at org.tigris.subversion.subclipse.ui.operations.SVNOperation.run
(SVNOperation.java:90)
        at org.eclipse.team.internal.ui.actions.JobRunnableContext.run
(JobRunnableContext.java:146)
        at org.eclipse.team.internal.ui.actions.JobRunnableContext
$ResourceJob.runInWorkspace(JobRunnableContext.java:72)
        at org.eclipse.core.internal.resources.InternalWorkspaceJob.run
(InternalWorkspaceJob.java:38)
        at org.eclipse.core.internal.jobs.Worker.run(Worker.java:76)

What's interesting about this is that this is the exception that one
of my co-workers was getting "svn: Malformed network data" for a full
day and could not connect to the SVN server over SSH at all. The
problem went away when he upgraded his network drivers, oddly enough,
but still I think this might be the culprit as this exception is
thrown every 50 lines or so.

Bradley

On May 4, 2006, at 3:10 PM, Alexander Kitaev wrote:

> Hello,
>
>> Not sure if this speaks to your point about symbolic linking
>> but I tried checking out "trunk" instead of a branch and the
>> time discrepancy was substantial. The checkout of the trunk
>> happened in about 6 minutes instead of 12, but was still
>> minutes slower than the same checkout over JavaHL.
> Thanks for reporting that. This difference could be only explained
> by the
> server-side delays (for the client there is no difference in trunk
> or branch
> checkout in case amount of files is more or less the same).
>
> So, please get a debug log as I described - it will help to
> understand how
> much time is spent on the server and how much is on the client
> side. I will
> probably have more questions/tests for you after I'll get this log.
> Thanks!
>
> Alexander Kitaev,
> TMate Software,
> http://tmate.org/
> http://jetbrains.com/tmate/
>
>> -----Original Message-----
>> From: Bradley Wagner [mailto:bradley.wagner@hannonhill.com]
>> Sent: Thursday, May 04, 2006 21:07
>> To: users@subclipse.tigris.org
>> Subject: Re: [Subclipse-users] Slowness in svn+ssh
>>
>> Alexander,
>> Not sure if this speaks to your point about symbolic linking
>> but I tried checking out "trunk" instead of a branch and the
>> time discrepancy was substantial. The checkout of the trunk
>> happened in about 6 minutes instead of 12, but was still
>> minutes slower than the same checkout over JavaHL.
>>
>> Then, I popped in your javasvn.jar into the plugin's lib
>> directory and tried again. The results were about the same
>> though (12m for branch, 6m for trunk).
>>
>> Bradley
>>
>> On May 4, 2006, at 2:40 PM, Alexander Kitaev wrote:
>>
>>> Hello Bradley,
>>>
>>>> Mark, I'm REALLY glad you brought this up. I tried this using the
>>>> svn:// protocol and the checkout ran in 1m45s over JavaHL and in
>>>> about 12m over JavaSVN. So, I would say there is a fundamental
>>>> problem with JavaSVN the way it's running now. I also
>> notice that my
>>>> whole computer slowed down when performing the checkout under
>>>> JavaSVN.
>>> I just would like to add that it is not the way JavaSVN
>> usually works,
>>> probably there is something specific in your environment that makes
>>> JavaSVN slow. I'm not saying that there are no bugs or performance
>>> problems in JavaSVN or that I'm not going to fix this very problem,
>>> but for most of the users it works more or less fast as far
>> as I know.
>>>
>>> Did you tried the version I sent you recently?
>>>
>>> Alexander Kitaev,
>>> TMate Software,
>>> http://tmate.org/
>>> http://jetbrains.com/tmate/
>>>
>>>> -----Original Message-----
>>>> From: Bradley Wagner [mailto:bradley.wagner@hannonhill.com]
>>>> Sent: Thursday, May 04, 2006 20:37
>>>> To: users@subclipse.tigris.org
>>>> Subject: Re: [Subclipse-users] Slowness in svn+ssh
>>>>
>>>> Mark, I'm REALLY glad you brought this up. I tried this using the
>>>> svn:// protocol and the checkout ran in 1m45s over JavaHL and in
>>>> about 12m over JavaSVN. So, I would say there is a fundamental
>>>> problem with JavaSVN the way it's running now. I also
>> notice that my
>>>> whole computer slowed down when performing the checkout under
>>>> JavaSVN.
>>>>
>>>> I am using Eclipse 3.1.2 on a Mac OS X client machine. I installed
>>>> Java 5 Release 4 for the Mac recently which makes itself
>> the default
>>>> meaning that Eclipse itself is running under Java 1.5.
>>>>
>>>> Next I tried blowing away my Eclipse install and starting
>> fresh with
>>>> 3.1.2 and Subclipse 1.0.1 and running the same test over
>> svn:// and
>>>> the results of the project checkout were about the same.
>>>>
>>>> I got someone else to connect to the same server from Windows XP
>>>> machine and run the same tests. They got similar results.
>>>> JavaHL performed the checkout in 2 minutes, JavaSVN in about 12.
>>>>
>>>> Bradley
>>>>
>>>> On May 4, 2006, at 12:00 PM, Mark Phippard wrote:
>>>>
>>>>> Bradley Wagner <bradley.wagner@hannonhill.com> wrote on 05/04/2006
>>>>> 11:58:05 AM:
>>>>>
>>>>>>> Since you have it local, why not try it using svn://?
>>>>>>
>>>>>> I certainly will try it to test it out. Unfortunately, it's not
>>>>>> feasible for everyone connecting to the repository to do it with
>>>>>> svn:// as we have some firewall issues.
>>>>>
>>>>> Yeah, I only meant as a test. I would expect JavaHL and
>> JavaSVN to
>>>>> perform real similar via svn://. If the same time difference was
>>>>> present it would seem like something was really amiss.
>>>>>
>>>>> Mark
>>>>>
>>>>>
>>>>>
>>>>
>> _____________________________________________________________________
>>>> _
>>>>> _______
>>>>> Scanned for SoftLanding Systems, Inc. and SoftLanding
>> Europe Plc by
>>>>> IBM Email Security Management Services powered by MessageLabs.
>>>>>
>>>>
>> _____________________________________________________________________
>>>> _
>>>>> _______
>>>>>
>>>>>
>>>>
>> ---------------------------------------------------------------------
>>>>> To unsubscribe, e-mail: users-unsubscribe@subclipse.tigris.org
>>>>> For additional commands, e-mail: users-help@subclipse.tigris.org
>>>>>
>>>>
>>>>
>>>> --
>>>> Bradley Mitchell Wagner
>>>> Software Developer
>>>> Hannon Hill Corporation
>>>> main: (678) 904-6900 ext. 115
>>>> email: bradley.wagner@hannonhill.com
>>>>
>>>>
>>>>
>>>>
>> ---------------------------------------------------------------------
>>>> To unsubscribe, e-mail: users-unsubscribe@subclipse.tigris.org
>>>> For additional commands, e-mail: users-help@subclipse.tigris.org
>>>>
>>>>
>>>
>>>
>>>
>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: users-unsubscribe@subclipse.tigris.org
>>> For additional commands, e-mail: users-help@subclipse.tigris.org
>>
>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@subclipse.tigris.org
>> For additional commands, e-mail: users-help@subclipse.tigris.org
>>
>>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subclipse.tigris.org
> For additional commands, e-mail: users-help@subclipse.tigris.org
>

--
Bradley Mitchell Wagner
Software Developer
Hannon Hill Corporation
main: (678) 904-6900 ext. 115
email: bradley.wagner@hannonhill.com
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subclipse.tigris.org
For additional commands, e-mail: users-help@subclipse.tigris.org
Received on Thu May 4 22:48:43 2006

This is an archived mail posted to the Subclipse Users mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.