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

odd (corrupted?) revision file header

From: <kmradke_at_rockwellcollins.com>
Date: Tue, 28 Jan 2014 16:08:22 -0600

I'm attempting to diagnose the cause of a corrupted revision file. It was
created with mod_dav_svn 1.7.5 and has the following top few lines:

0000000: 4445 4c54 410a 0000 0004 0200 0012 5356 DELTA.........SV
0000010: 4e01 0000 86a0 0005 8498 3404 8086 a000 N.........4.....
0000020: 86a0 0078 5eec ba65 501d 5db4 267c 08ee ...x^..eP.].&|..

As far as I read the libsvn_fs_fs/structure and notes/svndiff documents,
this isn't valid because it has some binary data between the DELTA
and SVN tokens.

svnadmin verify does indeed flag this revision as:

svnadmin: E160004: Corrupt representation '1081 0 80306 178176 [...]
svnadmin: E160004: Malformed svndiff data in representation

The svndiff data for the other added files in the revision is valid and
occurs after this part of the file. The other files in the revision
are exportable and valid. (There were 6 total added files and 5
deleted files in this revision. The revision is 138MB in size.)

This does not appear to be disk corruption since the file is
identical to backups of when it was originally created.
Unfortunately it is 19 revs outside of my transaction backup
dump window but I suspect that post-commit svnadmin dump
failed anyway.

fsfsverify.py doesn't seem to address this particular issue.

Thoughts on what could have caused the extra bytes?

Kevin R.
Received on 2014-01-28 23:11:01 CET

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