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

Re: endless(?) loop with heavy io when doing a svn move

From: AmenophisIII <AmenophisIII_at_gmx.at>
Date: Fri, 6 Feb 2009 07:45:07 +0100

On Thu, 5 Feb 2009 22:52:44 -0600
Ryan Schmidt <subversion-2009a_at_ryandesign.com> wrote:

> > at least the url of the rep is in there, which i
> > dont wanna share anyway.
> > can i get rid of that information without killing the bug trigger?
>
> I'm not sure what's causing the bug, but you can use "svn switch --
> relocate your_private_url some_other_url" to get the URL out of the
> working copy. Then you can test whether you can still reproduce the bug.

thx for the fast reply!
the switch --relocate command requires a working repo somewhere else
(it tries to connect there etc.)
i edited the entries file per hand and the bug still triggers.
the directory tar is attached.
a "svn mv test.vhd lala.vhd" should trigger the bug.
i observed three different behaviors dependent upon when you
send the sigkill (which ive done with ctrl+c all the time):

1. if you are fast enough, it will stop without a lot of diskio (maybe
even read only) after a few seconds of high cpu load.
2. a little bit longer and will recover after a minute or so (with io).
i _think_ i did not have to cleanup the dir afterwards.
3. if you ctrl+c later than 3-4 seconds after executing, it will trash
the harddisk for several minutes and you have to svn cleanup.

as i said, a commit fixes the problem.
if you need any other information, just mail me (e.g. for filing an
actual bug report).
but im not very interested in testing this further and didnt even
dare to strace it, because it kills my virtualbox VMs running in the
background :)

greets and thx for the (otherwise ;) good work!

Received on 2009-02-06 07:48:09 CET

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