You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Thomas Beale <th...@deepthought.com.au> on 2006/02/01 08:13:06 UTC

corrupted FSFS repository

Hi

I have spent a few days looking for evidence and reports of this bug, 
which I posted on the users list. It appears to be known as issue 
2467...http://subversion.tigris.org/issues/show_bug.cgi?id=2467

Our experience of the problem is the same as reported in the issue.

At the moment I can see no guidance on how to fix or even diagnose this. 
I cannot see how to find out which file is the offender. There seems to 
be no repair facility for fsfs repositories.

As far as I can see I will have to recreate the repository. But what 
will stop this happening again? No-one seems to know what causes it. (my 
only surmise is that it is to do with binary files, as we have a lot of 
those - mainly PDF and FrameMaker).

I don't want to sound alarmist, but this seems like a serious bug, or am 
I missing something?

this is what I originally reported.........................

Our experience of this problem is: doing a new checkout or an update to 
one of our repositories reslts in a partial set of files, then an error:

Error: REPORT request failed on '/specification/!svn/vcc/default'
Error: The REPORT request returned invalid XML in the response: XML 
parse error at
Error: line 1888829: no element found (/specification/!svn/vcc/default)

Multiple users have the same experience. No-one can successfully 
checkout or update (both fail aprtway through). The environment is:

Server:
Debian Linux 3.0
Subversion 1.2.0 (r14790) (compiled june 29 2005)
file system - FSFS
Apache 2.0.54

Clients:
generally windows XP + near-latest Tortoise

If I go onto the server and do a svnadmin verify, I get:

....
* Verified revision 47.
* Verified revision 48.
* Verified revision 49.
* Verified revision 50.
svnadmin: Can't read length line in file 'specification/db/revs/51'

It seems as if the last check-in has broken the repository. In it there 
is a PDF file (which has been checked in many times before; and there 
are many PDFs in our repo) that when the file name is clicked in the web 
view (or wsvn), you get:

<D:error>
<C:error/>
     <m:human-readable errcode="200002">
Can't read length line in file '/usr/local/var/svn/specification/db/revs/51'
</m:human-readable>
</D:error>

But clicking on other files in the revision seems to be ok. I don't 
understand this message: does it mean the actual file I just requested 
(happens to be called adl.pdf), or something else?

If anyone is interested, you can try the links at 
http://svn.openehr.org/specification/BRANCHES/Release-1.0-candidate/publishing/index.html, 
the repo root is http://svn.openehr.org/specification. The link 
http://svn.openehr.org/specification/BRANCHES/Release-1.0-candidate/publishing/architecture/am/adl.pdf 
is the broken one.

- thomas beale




---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by Rick Tessner <ri...@onnadayr.ca>.
Hi all,

On Wed, 2006-08-02 at 05:31 -0500, John Szakmeister wrote:
> On Sunday 05 February 2006 10:55, Marc Sherman wrote:
> > John Szakmeister wrote:
> > > If you have the original repository handy, this is actually very easy
> > > to fix. You can send me a private email if you need that done, but it
> > > looks like you've already gotten back up and running using a
> > > different solution.
> >
> > John, do you or Malcome have a tool (or instructions) for repairing
> > these corrupted repositories that's publicly available?  I've been
> > living in fear of this bug hitting the repositories I manage at work for
> > months now. :)
> 
> Have no fear.  It's at least possible to recover with no data loss (at least 
> that's what I've experienced with the ones that I have personally fixed... 
> after I found the magic recipe).
> 
> I've attached my script.  Your probing motivated me to automate the fix of 
> this problem. :-)

In less than a month, I've now been hit twice with a corrupted
repository.  First the techy bits:

        Platform: Linux, Ubuntu Breezy (2.6.12-10-686-smp).
        Subversion: 1.3.0
        Access via: https / mod_dav_svn
        Subversion store: FSFS
        Disk mirroring on the repository, no errors reported.

The corruption has come from the same user, Windows XP, TortoiseSVN
1.3.1.

I cannot, unfortunately, give access to the repos or the rev causing the
grief.  The best I can do is try to summarize what one of these commits
looks like:

      * very large commits (latest 83 MB, previous 400 MB)
      * with a mix of large & small files
      * upwards of 120 files & folders
      * long pathnames (up to 169 en-US characters)
      * mix of characters (including spaces, parens, tildes)
      * a variety of file formats: PDF, CAD diagrams, zip files, Word
        docs, Excel spreadsheets
      * slowish network connection
      * other folk with faster network connection committing at same
        time.

I've tried the fsfsverify.py script (provided kindly by John Szakmeister
to this list back on Feb 8, 2006) with the following results (both
times, same error).

Results of svnadmin verify repos
---------------------------------
...
* Verified revision 3183.
svnadmin: Unexpected end of svndiff input

Results of svnadmin dump --incremental -r 3184 repos > /tmp/x
----------------------------------------------------------------
svnadmin: Unexpected end of svndiff input 

        I do have a post-commit hook that does an incremental dump and
        the results of that are identical to the results of the dump
        done above.  So, I guess that means that the corruption does
        occur during the commit process and not something after that
        commit has occurred.

Results of fsfsverify.py --fix-read-length-line-error repos/db/revs/3184
------------------------------------------------------------------------

Traceback (most recent call last):
  File "../fsfsverify.py", line 683, in ?
    process(noderev, rev_file, options.dump_instructions,
options.dump_windows)
  File "../fsfsverify.py", line 636, in verify
    dump_windows)
  File "../fsfsverify.py", line 273, in verify
    digest = parse_svndiff(f, self.length, dump_instructions,
dump_windows)
  File "../fsfsverify.py", line 162, in parse_svndiff
    (instr_size, data_length, target_length_tmp) = \
  File "../fsfsverify.py", line 82, in parse_svndiff_instruction
    raise "Invalid instruction found at offset %d" % start
Invalid instruction found at offset 81484719

I'd absolutely love to help in getting this solved.  If there's anything
else that I can provide, please let me know.

-- 
rick@onnadayr.ca
IT Firefighting and Prevention


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by John Szakmeister <jo...@szakmeister.net>.
On Sunday 05 February 2006 10:55, Marc Sherman wrote:
> John Szakmeister wrote:
> > If you have the original repository handy, this is actually very easy
> > to fix. You can send me a private email if you need that done, but it
> > looks like you've already gotten back up and running using a
> > different solution.
>
> John, do you or Malcome have a tool (or instructions) for repairing
> these corrupted repositories that's publicly available?  I've been
> living in fear of this bug hitting the repositories I manage at work for
> months now. :)

Have no fear.  It's at least possible to recover with no data loss (at least 
that's what I've experienced with the ones that I have personally fixed... 
after I found the magic recipe).

I've attached my script.  Your probing motivated me to automate the fix of 
this problem. :-)

*** DISCLAIMER ***
THERE IS NO WARRANTY PROVIDED.  USE AT YOUR OWN RISK!

I originally developed this tool to help locate the problem and examine some 
of the metadata surrounding it so that I didn't have to look at a user's 
proprietary data.  It was never geared toward fixing this problem, and it was 
never meant to be a real application of any sort.  It's a complete hack.  
However, I've used this on several bad rev files and it has successfully 
restored them to full health.

First, make a backup of your repository!  The tool is meant to be run against 
the bad rev file.  Running 'fsfsverify <rev-file>' will attempt to verify the 
contents and notify you of the problem.  It dumps a ton of other 
information... that's normal.  To try and fix the file, run 'fsfsverify -f 
<rev-file>'.  It runs through the same process as the previous command, but 
this time it catch the error and attempt to fix the problem.  It *will* 
modify the rev file, so it's very important that you keep a backup copy of 
the file.  After everything is fixed, you should do a dump and load of the 
repository to remove the extra cruft that's in there.  If you don't care 
about it, make sure to at least run 'svnadmin verify' to make sure everything 
is okay.  On that note, let me just say once again, that *everyone* should be 
running 'svnadmin verify' or 'svnadmin dump' as part of a backup process to 
verify the integrity of their repository.

If you still have a problem afterwards, contact me.  I may be able to fix it, 
if I have access to the repository.

> Also, what's the current timeline for fixing this problem in subversion?
>  Is it expected to go into a 1.3.x release, or will it have to wait for
> 1.4.0?

Honestly, I don't know.  I'm sorry to say that my time for Subversion has been 
very limited lately. :-(  I believe mod_dav_svn is the source of the problem 
in that it's somehow opening two handles to the rev file.  Unfortunately, 
this problem is so intermittent, that I've been unable to reproduce it.  
Malcome has proposed some changes to the FSFS backend that would prevent two 
handles from being opened, but I'm not sure there was consensus on them.  At 
this point we aren't certain which line of code is in error, so it's hard to 
say when a proper fix will be available.

I will say that 1.4.0 will introduce a new svndiff algorithm.  Repositories 
created with it will not verify correctly with this script.  At this time, I 
don't plan on updating the script unless I run into the problem, or I see 
enough corruptions to motivate me to do it. :-)

-John

Re: corrupted FSFS repository

Posted by Thomas Beale <th...@deepthought.com.au>.
just a few possibly helpful details....

> 
> With that in mind, some mitigating factors/guesses:
> 
> * We think this can only happen if you get something like a write error
> while writing the file.  In about half(?) of the cases, users subsequently
> reported disk problems - so if your disk is okay, you're probably less
> likely to hit this.

no disk problem for us.

> 
> * Possibly coupled with the previous, versions of APR prior to 0.9.7
> didn't report write errors on buffered files at all - if you're using
> APR 0.9.7, you might be in a better position.
> 
> * I think every report has been for a server using mod_dav_svn.
> It's possible that svnserve isn't affected (or then again, that it's
> just less commonly used).

my instance of this bug is mod_dav_svn.

> 
> * Finally, if we are overwriting the revfile after we commit the
> transaction, it's possible that this is only a problem on non-Windows
> servers, since I don't think you can rename an open file on Windows.

and our server is debian.

- thomas beale


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by Marc Sherman <ms...@projectile.ca>.
Malcolm Rowe wrote:
> 
> I think 'living in fear' is a bit strong: we've had, what, five/six
> reports? ;)  

I have a very healthy sense of paranoia.  It's worth noting that I'm
running a linux server with mod_dav_svn, so I am in the target market
for this bug.

John Szakmeister wrote:
> 
> Have no fear.  It's at least possible to recover with no data loss (at least 
> that's what I've experienced with the ones that I have personally fixed... 
> after I found the magic recipe).
> 
> I've attached my script.  Your probing motivated me to automate the fix of 
> this problem. :-)

Thanks very much.  I'll stash this away on my server for later use.  You
might want to get this checked in to contrib.

- Marc





---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by Malcolm Rowe <ma...@farside.org.uk>.
On Wed, Feb 08, 2006 at 02:49:06PM -0500, Daniel Berlin wrote:
> On Wed, 2006-02-08 at 16:18 +0100, Folker Schamel wrote:
> > Why isn't it the only valid behaviour that the transaction is cancelled
> > if there is a disk write error?
> It is.
> 
> Note that this is why it is buggy :)
> 
> We're just not sure *where* this is happening.
> 

I think Folker may have been asking why FSFS doesn't enforce that
behaviour by marking the transaction as doomed when it sees a disk write
error, rather than assuming that the caller will do it.  And he's right,
it should.

The main reasons I hadn't gone down that path already are (as I wrote
originally): a) that APR <= 0.9.6 doesn't report those errors anyway,
and b) that I thought that adding code to check explicitly for EIO on
every write would be unacceptably unmaintainable.

However, now I come to think of it, the approach I have been thinking
about (to ensure that the file handles don't remain open on error)
should be extensible to killing the transaction on error as well,
without a great deal of difficulty or extra invasiveness.

(Another problem I've just noticed is that APR doesn't actually seem to
provide any way to check for EIO returns, so I think we'd probably have
to just treat any 'unexpected' error as a dooming error.)

If I do get around to doing the former, I'll certainly look at the latter
as well.

Regards,
Malcolm

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by Daniel Berlin <db...@dberlin.org>.
On Wed, 2006-02-08 at 16:18 +0100, Folker Schamel wrote:
> Malcolm Rowe wrote:
> 
> > ...
> > 
> > Disk: 1=================
> > Mem:                    ...
> > 
> > For whatever reason, we don't abort the transaction, but instead write
> > out the file's contents again.  This succeeds, so now we have this:
> > 
> > Disk: 1=================2########################>
> > Mem:                    ...
> > 
> > ...
> > 
> > It's not currently planned for a release, since we're not entirely sure
> > how best to fix it.  There's a couple of approaches:
>  >
>  > ...
> 
> Why isn't it the only valid behaviour that the transaction is cancelled
> if there is a disk write error?
It is.

Note that this is why it is buggy :)

We're just not sure *where* this is happening.



---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by Folker Schamel <sc...@spinor.com>.
Malcolm Rowe wrote:

> ...
> 
> Disk: 1=================
> Mem:                    ...
> 
> For whatever reason, we don't abort the transaction, but instead write
> out the file's contents again.  This succeeds, so now we have this:
> 
> Disk: 1=================2########################>
> Mem:                    ...
> 
> ...
> 
> It's not currently planned for a release, since we're not entirely sure
> how best to fix it.  There's a couple of approaches:
 >
 > ...

Why isn't it the only valid behaviour that the transaction is cancelled
if there is a disk write error?
Otherwise the repository contains broken file content.

Cheers,
Folker


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by Malcolm Rowe <ma...@farside.org.uk>.
On Sun, Feb 05, 2006 at 10:55:42AM -0500, Marc Sherman wrote:
> John, do you or Malcome have a tool (or instructions) for repairing
> these corrupted repositories that's publicly available?  I've been
> living in fear of this bug hitting the repositories I manage at work for
> months now. :)
> 

I think 'living in fear' is a bit strong: we've had, what, five/six
reports? ;)  Maybe I should explain what I think is happening:

We start off by writing the rev-file for a particular file that's
changed.  We hit an error, and so part of the file's contents is left
in an in-memory buffer.  The file looks like this:

Disk: 1=================
Mem:                    ...

For whatever reason, we don't abort the transaction, but instead write
out the file's contents again.  This succeeds, so now we have this:

Disk: 1=================2########################>
Mem:                    ...

Note that we still have the in-memory buffer from the first file open.
The revfile now has some crud at the front, but it's otherwise perfectly
valid.  The problem comes when we finally destroy the pool that the
first filehandle was in: it flushes its buffered contents to the revfile
(this might even be after we've committed the transaction).

Now we have this:
Disk: 1=================...######################>

The start of the second noderev (the successfully written one) has been
overwritten by the middle of the second, corrupting it.  Fortunately,
the fix is simple: since the two noderevs are identical, we can copy&paste
the start of the second noderev from earlier:

Disk: 1=================1==######################>

And now the noderev is valid again.  This is what John's tool does, IIRC.


With that in mind, some mitigating factors/guesses:

* We think this can only happen if you get something like a write error
while writing the file.  In about half(?) of the cases, users subsequently
reported disk problems - so if your disk is okay, you're probably less
likely to hit this.

* Possibly coupled with the previous, versions of APR prior to 0.9.7
didn't report write errors on buffered files at all - if you're using
APR 0.9.7, you might be in a better position.

* I think every report has been for a server using mod_dav_svn.
It's possible that svnserve isn't affected (or then again, that it's
just less commonly used).

* Finally, if we are overwriting the revfile after we commit the
transaction, it's possible that this is only a problem on non-Windows
servers, since I don't think you can rename an open file on Windows.

The biggest problem is that we've just not been able to reproduce this,
so we can't see exactly what's going on.


> Also, what's the current timeline for fixing this problem in subversion?
>  Is it expected to go into a 1.3.x release, or will it have to wait for
> 1.4.0?
> 

It's not currently planned for a release, since we're not entirely sure
how best to fix it.  There's a couple of approaches:

1. Fix the problem at source (i.e. mod_dav_svn, if it's that).
Trouble is, we aren't really sure how to proceed with tracking down the
source, so this is a bit of a dead end.  This also doesn't fix any other
badly-behaved clients.

2. Kill the FSFS transaction if we spot a fatal error.  Except that we
have to distinguish 'fatal' from 'non-fatal' errors, and as previously
mentioned, APR 0.9.6 and below don't report errors writing buffered
files at all.  (and the error-handling mechanics would likely be quite
intrusive).

3. Make sure FSFS uses only one file handle for the transaction's revfile.
This wouldn't be a bad idea anyway, since we currently do a lot of
open/seek-to-end/close throughout a transaction.  The problem here is
that there's no unique 'transaction' object that a client has to use to
modify a transaction (it can legitimately use two 'transaction handle'
objects opened from different 'filesystem handle' objects to update the
same transaction), so we'd have to maintain some kind of synchronised
reference-counted hash of {filesystem UUID, txn id} => {transaction
object}, which is quite an overhead just for this.

4. Finally, the idea I'm currently favouring: Make sure that everywhere
we open a file for write in FSFS, we ensure that it's closed before we
return.  This is simply a matter of calling apr_file_close() everywhere,
including in our error paths.  It's likely to be fairly intrusive (for
libsvn_fs_fs/fs_fs.c), but not too complex, I hope.

Regards,
Malcolm

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by Marc Sherman <ms...@projectile.ca>.
John Szakmeister wrote:
> 
> If you have the original repository handy, this is actually very easy
> to fix. You can send me a private email if you need that done, but it
> looks like you've already gotten back up and running using a
> different solution.

John, do you or Malcome have a tool (or instructions) for repairing
these corrupted repositories that's publicly available?  I've been
living in fear of this bug hitting the repositories I manage at work for
months now. :)

Also, what's the current timeline for fixing this problem in subversion?
 Is it expected to go into a 1.3.x release, or will it have to wait for
1.4.0?

Thanks for the info,
- Marc

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by Thomas Beale <Th...@OceanInformatics.biz>.
John Szakmeister wrote:
> On Wednesday 01 February 2006 03:13, Thomas Beale wrote:
> [snip]
>   
>> If I go onto the server and do a svnadmin verify, I get:
>>
>> ....
>> * Verified revision 47.
>> * Verified revision 48.
>> * Verified revision 49.
>> * Verified revision 50.
>> svnadmin: Can't read length line in file 'specification/db/revs/51'
>>     
>
> This looks like a bug that Malcome Rowe and myself have been trying to track 
> down.  What appears to be happening, is that an alternate handle to the rev 
> file is being opened and used, while the other is left dangling.  When pool 
> cleanup time comes around, it flushes the remaining data in the buffer, 
> clobbering one of the text representations.
>
> If you have the original repository handy, this is actually very easy to fix.  
> You can send me a private email if you need that done, but it looks like 
> you've already gotten back up and running using a different solution.
>   
well, I wouldn't call it a solution - I used sledge hammer means to just 
toss the "corrupted" most recent revision, and then more or less 
reconstructed it from my workarea (which meant recreating that as well, 
since it lost any coherent idea of what state the repository is in). In 
practical terms, everything is back up and working, with some annoyance 
factor - everyone with a copy of that repo had to do a new clean checkout.

But I would be interested to know the proper "fix" that actually 
uncorrupts the repository - can you publish that? My researches on lists 
found about 5 or so instances of this bug being experienced 
independently, so it is a real one. Even a recovery fix published on the 
list (or better in the FAQ) would be better than nothing - eventually I 
imagine someone will find the offending bug in the code, and then we 
will all be happy;-)

- thomas beale

> -John
>
>
>   


-- 
___________________________________________________________________________________
CTO Ocean Informatics (http://www.OceanInformatics.biz)
Research Fellow, University College London (http://www.chime.ucl.ac.uk)
Chair Architectural Review Board, openEHR (http://www.openEHR.org)


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by John Szakmeister <jo...@szakmeister.net>.
On Wednesday 01 February 2006 03:13, Thomas Beale wrote:
[snip]
> If I go onto the server and do a svnadmin verify, I get:
>
> ....
> * Verified revision 47.
> * Verified revision 48.
> * Verified revision 49.
> * Verified revision 50.
> svnadmin: Can't read length line in file 'specification/db/revs/51'

This looks like a bug that Malcome Rowe and myself have been trying to track 
down.  What appears to be happening, is that an alternate handle to the rev 
file is being opened and used, while the other is left dangling.  When pool 
cleanup time comes around, it flushes the remaining data in the buffer, 
clobbering one of the text representations.

If you have the original repository handy, this is actually very easy to fix.  
You can send me a private email if you need that done, but it looks like 
you've already gotten back up and running using a different solution.

-John

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by Thomas Beale <th...@deepthought.com.au>.
Bart Robinson wrote:
> On 2006-2-2 Thomas Beale <th...@deepthought.com.au> wrote:

> 
> So it seems that removing the rev file is a workaround.
> Did you also have to remove the revprop file or edit any other
> files in the repository?

Michael Brouwer told me to do the following (which worked) - this is to 
remove the Nth revision of a repository:

- delete revision N from db/revs
- delete revision N from db/revprops
- change the first field in db/current back to N-1

(of course you probably want to turn off Apache and any local subversion 
access before doing this).

I would have thought that there should be a script that does this....

- thomas beale


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by Bart Robinson <lo...@pobox.com>.
On 2006-2-2 Thomas Beale <th...@deepthought.com.au> wrote:
 > Thomas Beale wrote:
 > > 
 > > Hi
 > > 
 > > I have spent a few days looking for evidence and reports of this bug, 
 > > which I posted on the users list. It appears to be known as issue 
 > > 2467...http://subversion.tigris.org/issues/show_bug.cgi?id=2467
 > > 
 > > Our experience of the problem is the same as reported in the issue.
 > > 
 > 
 > I made an attempt to reproduce this as follows:
 > 
 > - removed revision 51 (the last & broken one) from the repository by hand
 > - recreated my workarea in such a way as to look (nearly) like it would 
 > have prior to the commit that caused the broken revision
 > - did the same commit (can't guarantee the same byte for byte, but the 
 > PDF file which returned the XML error was in fact unchanged).

 > This commit worked fine, and subsequent commits did as well. Good for us 
 > in that the repo is working again, but bad in that I couldn't diagnose 
 > anything - all I can say is that it discounts anything in the particular 
 > PDF file. This would point more to a network problem, apache problem or 
 > some other non content-related problem occurring during a commit.

So it seems that removing the rev file is a workaround.
Did you also have to remove the revprop file or edit any other
files in the repository?

-- bart

 > If we get the error again, I will make further efforts to reproduce and 
 > report it here.
 > 
 > - thomas beale

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: corrupted FSFS repository

Posted by Thomas Beale <th...@deepthought.com.au>.
Thomas Beale wrote:
> 
> Hi
> 
> I have spent a few days looking for evidence and reports of this bug, 
> which I posted on the users list. It appears to be known as issue 
> 2467...http://subversion.tigris.org/issues/show_bug.cgi?id=2467
> 
> Our experience of the problem is the same as reported in the issue.
> 

I made an attempt to reproduce this as follows:

- removed revision 51 (the last & broken one) from the repository by hand
- recreated my workarea in such a way as to look (nearly) like it would 
have prior to the commit that caused the broken revision
- did the same commit (can't guarantee the same byte for byte, but the 
PDF file which returned the XML error was in fact unchanged).

This commit worked fine, and subsequent commits did as well. Good for us 
in that the repo is working again, but bad in that I couldn't diagnose 
anything - all I can say is that it discounts anything in the particular 
PDF file. This would point more to a network problem, apache problem or 
some other non content-related problem occurring during a commit.

If we get the error again, I will make further efforts to reproduce and 
report it here.


- thomas beale


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org