Betr: Re: Many missing revlogs, do we have a problem?

[Alban Hertroys]

Matt Mackall <mpm at selenic.com> wrote on 17/06/2015 18:57:08:

> 

alban.hertroys at apollovredestein.com
T: 




 	Apollo Vredestein B.V. - P.O. Box 27 - 7500 AA  Enschede - The Netherlands - 
Chamber of Commerce number 34223268 - http://www.apollovredestein.com	 

The information contained in this e-mail is intended solely for the use of the 
individual or entity to whom it is addressed and others authorized to receive 
it. You are hereby notified that any disclosure, copying, distribution or 
action in relation to the contents of this information is strictly prohibited. 
If you are not the intended recipient, please delete this message and any 
attachments and advise the sender by return e-mail. The confidentiality of this 
message is not warranted. Apollo Vredestein B.V. rules out any and every 
liability resulting from this or any other electronic transmission.
 
On Wed, 2015-06-17 at 16:58 +0200, Alban Hertroys wrote:
> > I just tried a synchronize with a local copy of our master repo and it 

> > failed with "empty or missing revlog". So I figured there might be a 
few 
> > corrupted revlogs somewhere and did an hg verify on our master repo 
(using 
> > tortoisehg 3.3). I was NOT prepared for what it told me, an excerpt 
from 
> > the end of the (long) list:
> > 
> > data/xxx_austrian_env/SalesreportPointSAustriaS.fex.i at 1883: missing 
> > revlog!
> >  data/xxx_austrian_env/SalesreportPointSAustriaS_RC.fex.i at 1883: 
missing 
> > revlog!
> >  data/xxx_austrian_env/SalesreportPointSAustriaStart.fex.i at 1883: 
missing 
> > revlog!
> >  data/xxx_austrian_env/points.csv.i at 1883: missing revlog!
> > 17922 files, 2741 changesets, 26977 total revisions
> > 13 warnings encountered!
> > 17985 integrity errors encountered!
> > (first damaged changeset appears to be 0)
> > 
> > It's telling me that we have even more integrity errors than we have 
> > files. Ouch!
> > 
> > Last time I synchronized to my local copy was sometime last week and 
that 
> > went fine. Now suddenly we're in _this_ situation, and of course just 
as 
> > we're about to release the committed code to our new production 
server. I 
> > find it rather hard to believe that hg verify is telling me the truth, 

> > frankly - it doesn't make sense...
> > 
> > Any idea how this might have happened?
> 
> Looks like someone/something deleted all the files on your server. Take
> a look at .hg/store/data on both the client and server.

A quick look ath the contents of .hg/store/data on both directories gives:

master    : 18,079 files, 356 directories, 64.5 MB size (123 MB on disk)
local copy: 17,929 files, 353 directories, 64.2 MB size (121 MB on disk)

That doesn't look like we've got a lot of deleted files there. Time to 
figure out a way to generate sorted lists of files on both, so we can find 
the differences.

I do notice a lot of .00changelog.i-* files in store/; I suppose those are 
commit message backups of failed commits (most would be due to the file 
locking issue)?

Also of note, more in respones to Simon's question: the fncache files are 
quite different in size between the two:
master    : 5 kB   (hmm... MS invented a new unit called 'KB', it seems)
local copy: 672 kB

> Fortunately every client repository is a complete copy. And of course,
> you also have nightly backups of your server, right?

Yes, we do indeed.

If we're going to use those though, we need a way to figure out how far 
back we need to go to fix this (I know _nothing_ of Windows backups). My 
local copy was last updated at June 12th, but we've been committing lots 
of changes the last couple of days in preparation for our new production 
server (which is already quite overdue due to issues with the new 
software), so the more recent the better.

Regards,

Alban.