Windows 10 random hg verify errors?

[Gregory Szorc]

On Tue, Feb 16, 2016 at 1:22 PM, Eric Pyle <eric_pyle at sil.org> wrote:

> *Has anyone* else experienced *random hg verify* errors under windows 10
> (64bit)?
>
> Basically, hg verify gives me different results on the same (albeit rather
> large 6Gigabyte) repository even though I haven't changed the repository
> (e.g. different files and count for "integrity errors encountered", or
> sometimes it succeeds without any errors.)
>
> *Does anyone* know of a similar large *public* project (with many files
> and a lot of history) that I could try to reproduce the error on that as
> well? (So I can invite support developers to try to reproduce it).
>
>
https://hg.mozilla.org/mozilla-central is pretty large. You can grab a
bundle at https://hg.cdn.mozilla.net/ and `hg unbundle` into an empty repo
to apply it.


> My current recipe for reproducing this problem: it seems to be more likely
> to occur soon after rebooting and logging into the system. Otherwise hg
> verify seems to be succeeding more often than not. As far as I can tell,
> the problem does not depend on the version of mercurial. I can reproduce
> random the errors on multiple versions.
>
> However, when I boot my system *under* *Safe-Mode* I *can't reproduce*
> these errors, random or otherwise. hg verify just passes like I expect.
> Likewise running hg verify from a windows 7 (64bit) machine on the same
> repository doesn't report such errors.
>
> I've reproduced the problem on both by cloning and booting a Solid State
> Drive (SSD) and non-SSD drive on the same system. Also, had a colleague
> with the same type of system/setup reproduce the problem by running hg
> verify on my repository via shared local network folder.
>
> Thanks for any feedback!
>
> Here's an example of two different runs (under TortoiseHg 3.7.1). As you
> can see by the timestamps, the first run reported 3 errors, but a
> subsequent run reported none.
>
> paratext% hg verify
> % hg verify
> checking changesets
> checking manifests
> crosschecking files in changesets and manifests
> checking files
>  Paratext/MainForm.Designer.cs at 17792: unpacking d73c194b3e16: integrity
> check failed on data/Paratext/MainForm.Designer.cs.i:229
>  PublishingAssistant/HelperClasses/CreateInDesignDocument4.cs at 22782:
> unpacking f45bc93b7f83: integrity check failed on
> data/PublishingAssistant/HelperClasses/CreateInDesignDocument4.cs.i:58
>  PublishingAssistant/NumberScripts/ChineseTraditionalLong.py at 10720:
> unpacking 368f3c4eadfc: integrity check failed on
> data/PublishingAssistant/NumberScripts/ChineseTraditionalLong.py.i:1
> 10518 files, 24071 changesets, 81121 total revisions
> 3 integrity errors encountered!
> (first damaged changeset appears to be 10720)
> [command returned code 1 Fri Feb 12 09:50:52 2016]
>
> paratext% hg verify
> % hg verify
> checking changesets
> checking manifests
> crosschecking files in changesets and manifests
> checking files
> 10518 files, 24071 changesets, 81121 total revisions
> [command completed successfully Fri Feb 12 09:53:44 2016]
>
> As I said, this particular "paratext" project repository is 6 Gigabytes.
> I'm guessing the large size of the repository in terms of project history
> and number of files (perhaps even filesize) increases the *likelihood* of
> reproducing this problem. Unfortunately my repository is private so I can't
> share it with others like mercurial support.
>
> *Here's a brief profile of my system:*
> Windows 10 Pro Version 1511 OS Build 10586.104
> Dell Optiplex 9020
> 16GB RAM
> i7-4790 CPU @ 3.60Ghz
> 64bit
>
> --
> Eric Pyle
> Dallas, TX
> SIL - Language Software Developer
>
> _______________________________________________
> Mercurial mailing list
> Mercurial at mercurial-scm.org
> https://www.mercurial-scm.org/mailman/listinfo/mercurial
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mercurial-scm.org/pipermail/mercurial/attachments/20160216/4f23531f/attachment-0002.html>