Broken repo? "abort: remotestore: largefile xxx is invalid"

[Scott Timothy E]

FWIW, I updated to the latest version of TortoiseHg / Mercurial and the update log now shows:

% hg update --repository C:\projects\gsx\fpga\jbtest --config ui.merge=internal:fail --rev 207 --clean
102 files updated, 0 files merged, 0 files removed, 0 files unresolved
getting changed largefiles
smartxplorer_results/run1/gsx_server_pb1b.pcf: data corruption (expected 2666200625286a694888ec6cd571fc779ebc7bd5, got a0e83c6556cb68a9ee346d7fa18d602d8ee95b79)
smartxplorer_results/run2/gsx_server_pb1b.pcf: data corruption (expected c024bcd2a2058369f993e5c291d2ed5c1981112f, got b43fe59586e6dcc4e4351d2428336b8e90f2d22c)
12 largefiles updated, 0 removed
[command completed successfully Tue Dec 18 11:28:46 2012]

So, the invalid condition is indicating hash differences I guess.  We had other tools open that could have potentially affected the contents of these files as the check-in was going on.  Does that sound like a likely scenario that would induce this sort of failure?

Thanks,
--tim


From: mercurial-bounces at selenic.com [mailto:mercurial-bounces at selenic.com] On Behalf Of Scott Timothy E
Sent: Tuesday, December 18, 2012 10:22 AM
To: mercurial at selenic.com
Subject: Broken repo? "abort: remotestore: largefile xxx is invalid"

I'm running into a problem using largefiles.  My google-fu has failed me, so I'm seeing if someone on this list might be able to lend some insight.

We've got a repo that has been using largefiles successfully for quite some time now.  Today, someone pushed something that broke things.  Here's the error we are seeing when trying to update to the revision with the problem:

% hg update --repository C:\projects\GSX\fpga\jbtest2 --config ui.merge=internal:fail --rev 208 --clean
380 files updated, 0 files merged, 178 files removed, 0 files unresolved
getting changed largefiles
abort: remotestore: largefile 2666200625286a694888ec6cd571fc779ebc7bd5 is invalid
[command returned code 255 Tue Dec 18 09:47:40 2012]
jbtest2%

If I manually copy the largefile in question from the main repo on the server to my local largefiles store, I can update successfully and things work fine.  So, I do not believe that the file itself is bad.  Does anyone know why the file may be detected as "invalid"?

The server machine is running WinXP.  My local machine is running Win7.  I'm using TortoiseHg 2.2.2 / Hg 2.0.2.  The person that apparently introduced the problem is running TortoiseHg 2.2.1 / Hg 2.0.1.

Thanks,
--tim

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mercurial-scm.org/pipermail/mercurial/attachments/20121218/b8aaa94f/attachment-0002.html>