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

Re: TortoiseSVN "Get lock" of multiple files fails via Java-SSL-tunnel since v1.9.0 (bug?)

From: Stefan Hett <stefan_at_egosoft.com>
Date: Tue, 10 May 2016 11:33:34 +0200

On 5/9/2016 3:20 PM, David Hinken wrote:
>> On 9 May 2016 at 15:23, David Hinken <d dot hinken at isfh dot de> wrote:
>>>>>> Thanks for suggesting these options. The SVN server is a visual svn server,
>>>>>> standard edition. I think that checking the logs and setting both options
>>>>>> might be difficult (?) but I will give it a try. I hope playing with these settings
>>>>>> does not have any side effects...
>>>>> VisualSVN Server always logs all errors to 'Application and Services'
>>>>> | 'VisualSVN Server' log in Event Viewer.
>>>> I will check that.
>>> The logs are full with 'Failed to create new lock. [423, #160035]' messages,
>>> followed by 'Path '/trunk/svn_test/992.txt' is already locked by user '...' in filesystem
>>> '...' [423, #160035]' messages... Not sure if this is the problem itself or just a
>>> follow-up because some of the files cannot be released (a 'break lock') is required.
>> This could happen for three reasons:
>> 1. The file is already locked before executing 'svn lock'.
>> 2. 'Java SSL Tunnel' mixes-up HTTP requests and send one LOCK request
>> multiple times.
>> 3. There is bug in ra_serf/serf and it sends the same requests more
>> than once in some circumstances.
>> I assume that you're using clean repository for tests, so (1) is not
>> likely the case. Bug in ra_serf/serf (3) is also possible, but I don't
>> think it's very likely since issue doesn't trigger when accessing
>> server directly. So in my opinion it's problem in Java SSL Tunnel (2).
>> HTTP proxies/filters known to be working really bad with HTTP
>> pipelining.
> Yes, a clean repository was used.
> If http proxies are known to work really bad with http pipelining it might be a nice svn-feature to switch pipelining on and off?
Given that SVN 1.9 is already out for several months and your report is
the first one which seems to be related to some http pipelining
problems, I wouldn't draw the conclusion that http proxies are in
general known to work bad with http pipelining. Do you have any other
reason to come to that conclusion?

Stefan Hett
To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_tortoisesvn.tigris.org].
Received on 2016-05-10 11:33:46 CEST

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

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