You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Karl Fogel <kf...@newton.ch.collab.net> on 2002/09/03 16:48:06 UTC
Re: Corrupted update
Blair Zajac <bl...@orcaware.com> writes:
> What can I do to help svn out when there's a corrupted update?
>
> I just updated svn to rev 3121 on my Windows 2000 box and the
> file subversion/libsvn_client/commit_util.c has some garbage
> in it starting at line 763.
Wow! Glad you noticed that, thanks.
The most helpful thing would be a reliable reproduction recipe. Do
you happen to know what revision(s) that tree was at before you
updated? Or even if not, can you recreate the problem by checking out
an older tree under Windows and then running the same update command
that gave you the corruption?
If we can get a repro recipe, we can fix it, guaranteed or your money
back :-).
-K
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Corrupted update
Posted by Blair Zajac <bl...@orcaware.com>.
Karl Fogel wrote:
>
> Blair Zajac <bl...@orcaware.com> writes:
> > > Is the file's text base is corrupted in the
> > > same way, or is it just the working file?
> >
> > They both have the identical md5sum.
>
> Okay, so their contents are identical.
>
> > Here are the entries from my Linux box and from the Windows box.
> > The checksum's differ:
>
> This could just be due to eol conversion stuff... (?)
The file does not have svn:eol-style set on it, so it shouldn't
have any eol conversions applied to it.
Best,
Blair
--
Blair Zajac <bl...@orcaware.com>
Web and OS performance plots - http://www.orcaware.com/orca/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Corrupted update
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Blair Zajac <bl...@orcaware.com> writes:
> > Is the file's text base is corrupted in the
> > same way, or is it just the working file?
>
> They both have the identical md5sum.
Okay, so their contents are identical.
> Here are the entries from my Linux box and from the Windows box.
> The checksum's differ:
This could just be due to eol conversion stuff... (?)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Corrupted update
Posted by Blair Zajac <bl...@orcaware.com>.
Karl Fogel wrote:
>
> Blair Zajac <bl...@orcaware.com> writes:
> > There was a Norton AntiVirus Enterprise scan going at the same time as
> > the upgrade, no telling if that had an effect.
> >
> > It looks like the repo was around 3095. I had another checked out svn
> > tree and tried the same upgrade, but that one worked, so it looks like
> > it was specific to this tree.
> >
> > Unfortunately, I didn't have a copy of the tree before the upgrade.
> >
> > The tree has just been upgraded over time, so coming up with a recipe
> > may be hard.
>
> Urk. Umm, okay, thanks anyway.
>
> What does 'svn st' say?
Nothing.
> Is the file's text base is corrupted in the
> same way, or is it just the working file?
They both have the identical md5sum.
> And if you run md5sum on
> the text base, does the result match the checksum given in the
> .svn/entries file?
I assume that the checksum in the .svn/entries is base64? I wrote
a little Perl script to give me the base64 md5sum of the file and
here's what I get:
Base 64 Md5sum
------------------------------
1B2M2Y8AsgTpgAmY7PhCfg commit_util.c
1B2M2Y8AsgTpgAmY7PhCfg .svn/text-base/commit_util.c.svn-base
EnE4v981FTIU+ImyWK0/GQ .svn/entries checksum for commut_util.c
Here are the entries from my Linux box and from the Windows box.
The checksum's differ:
Linux:
<entry
committed-rev="3113"
name="commit_util.c"
text-time="2002-09-03T16:55:19.000000Z"
committed-date="2002-09-01T23:09:53.884151Z"
checksum="1GIQ/STkAngwbfwujAYKnQ=="
last-author="philip"
kind="file"
prop-time="2002-09-03T16:55:19.000000Z"/>
Windows:
<entry
committed-rev="3113"
name="commit_util.c"
text-time="2002-09-03T16:48:36.246000Z"
committed-date="2002-09-01T23:09:53.884151Z"
checksum="EnE4v981FTIU+ImyWK0/GQ=="
last-author="philip"
kind="file"
prop-time="2002-09-03T16:48:35.964750Z"/>
Best,
Blair
--
Blair Zajac <bl...@orcaware.com>
Web and OS performance plots - http://www.orcaware.com/orca/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Corrupted update
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Blair Zajac <bl...@orcaware.com> writes:
> There was a Norton AntiVirus Enterprise scan going at the same time as
> the upgrade, no telling if that had an effect.
>
> It looks like the repo was around 3095. I had another checked out svn
> tree and tried the same upgrade, but that one worked, so it looks like
> it was specific to this tree.
>
> Unfortunately, I didn't have a copy of the tree before the upgrade.
>
> The tree has just been upgraded over time, so coming up with a recipe
> may be hard.
Urk. Umm, okay, thanks anyway.
What does 'svn st' say? Is the file's text base is corrupted in the
same way, or is it just the working file? And if you run md5sum on
the text base, does the result match the checksum given in the
.svn/entries file?
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Corrupted update
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
"Barry Scott" <ba...@ntlworld.com> writes:
> Have you thought of having svn keep a log of all the
> commands issued to help with recipe creation? We
> run this type of command line logging for our home grown
> SCM tools and its very useful in tracking down root
> cause of problems.
Yeah, but the thing is we usually also need the entire wc state from
before the command as well -- and that's expensive to save on every
command.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Corrupted update
Posted by Blair Zajac <bl...@orcaware.com>.
Barry Scott wrote:
>
> We see Norton AV locking files while its checking them out
> after you save a file to disk. It shows up when we save twice
> in quick succession.
>
> Have you thought of having svn keep a log of all the
> commands issued to help with recipe creation? We
> run this type of command line logging for our home grown
> SCM tools and its very useful in tracking down root
> cause of problems.
That's a good idea. Something like a shell script wrapper for svn
that logs each command into a file located in the top most directory
of the checked out tree and then runs the real command.
Do you have something like this already? If not, I'll hack something
up quickly.
Maybe if this is useful, then svn could do it itself. This would
be useful for us, because a lot of svn users are not going to keep
logs of this kind of stuff.
Best,
Blair
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
RE: Corrupted update
Posted by Barry Scott <ba...@ntlworld.com>.
We see Norton AV locking files while its checking them out
after you save a file to disk. It shows up when we save twice
in quick succession.
Have you thought of having svn keep a log of all the
commands issued to help with recipe creation? We
run this type of command line logging for our home grown
SCM tools and its very useful in tracking down root
cause of problems.
Barry
> -----Original Message-----
> From: Blair Zajac [mailto:blair@orcaware.com]
> Sent: 03 September 2002 18:43
> To: kfogel@collab.net
> Cc: Subversion Developers Mailing List
> Subject: Re: Corrupted update
>
>
> Karl Fogel wrote:
> >
> > Blair Zajac <bl...@orcaware.com> writes:
> > > What can I do to help svn out when there's a corrupted update?
> > >
> > > I just updated svn to rev 3121 on my Windows 2000 box and the
> > > file subversion/libsvn_client/commit_util.c has some garbage
> > > in it starting at line 763.
> >
> > Wow! Glad you noticed that, thanks.
> >
> > The most helpful thing would be a reliable reproduction recipe. Do
> > you happen to know what revision(s) that tree was at before you
> > updated? Or even if not, can you recreate the problem by checking out
> > an older tree under Windows and then running the same update command
> > that gave you the corruption?
> >
> > If we can get a repro recipe, we can fix it, guaranteed or your money
> > back :-).
>
> There was a Norton AntiVirus Enterprise scan going at the same time as
> the upgrade, no telling if that had an effect.
>
> It looks like the repo was around 3095. I had another checked out svn
> tree and tried the same upgrade, but that one worked, so it looks like
> it was specific to this tree.
>
> Unfortunately, I didn't have a copy of the tree before the upgrade.
>
> The tree has just been upgraded over time, so coming up with a recipe
> may be hard.
>
> Best,
> Blair
>
> --
> Blair Zajac <bl...@orcaware.com>
> Web and OS performance plots - http://www.orcaware.com/orca/
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: dev-help@subversion.tigris.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Corrupted update
Posted by Blair Zajac <bl...@orcaware.com>.
Karl Fogel wrote:
>
> Blair Zajac <bl...@orcaware.com> writes:
> > What can I do to help svn out when there's a corrupted update?
> >
> > I just updated svn to rev 3121 on my Windows 2000 box and the
> > file subversion/libsvn_client/commit_util.c has some garbage
> > in it starting at line 763.
>
> Wow! Glad you noticed that, thanks.
>
> The most helpful thing would be a reliable reproduction recipe. Do
> you happen to know what revision(s) that tree was at before you
> updated? Or even if not, can you recreate the problem by checking out
> an older tree under Windows and then running the same update command
> that gave you the corruption?
>
> If we can get a repro recipe, we can fix it, guaranteed or your money
> back :-).
There was a Norton AntiVirus Enterprise scan going at the same time as
the upgrade, no telling if that had an effect.
It looks like the repo was around 3095. I had another checked out svn
tree and tried the same upgrade, but that one worked, so it looks like
it was specific to this tree.
Unfortunately, I didn't have a copy of the tree before the upgrade.
The tree has just been upgraded over time, so coming up with a recipe
may be hard.
Best,
Blair
--
Blair Zajac <bl...@orcaware.com>
Web and OS performance plots - http://www.orcaware.com/orca/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org