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

Re: DB_INCOMPLETE: cache flush was unable to complete

From: Greg Goodrich <ggoodrich_at_medinotes.com>
Date: 2004-06-01 05:58:51 CEST

Alan Knowles wrote:

> I lost faith in DBD too - have a look at the development version
> there's a great filesystem called fsfs - good ole flat files...
>
> Regards
> Alan
>
>
> Erik Huelsmann wrote:
>
>>> I've been trying to migrate from VSS to Subversion. I've gotten my
>>> migration script working pretty well, but invariably, after so many
>>> "revisions" into Subversion, I get this error, and the whole thing
>>> shuts down. I then have to go to the SVN server and do a "svnadmin
>>> recover" on the db. Fortunately, I can restart my script where it
>>> left off (I finally got this ability somewhat added in), but it will
>>> run for a while longer and again throw this error.
>>>
>>> Quite frankly, I'm losing my faith in Subversion/BerkeleyDB. I was
>>> getting these errors in Subversion 1.0.2, so I've upgraded to 1.0.3,
>>> and now 1.0.4, and still I get these errors. I also updated my
>>> Berkeley DB to the latest version/patchlevel. I basically have to
>>> babysit the migration. I'm concerned that if/when I ever get our
>>> rather large VSS db converted, that we will have no end to the
>>> problems with the system. Does anyone out there have any ideas why
>>> these errors occur? It is somewhat consistent (only partially), as
>>> I've had it die out at revision 1108 about 15-20 times (prior to
>>> adding my restart ability). Of course the exact same files in the
>>> same order are being committed. However, sometimes it flies right
>>> by this revision and goes somewhat farther before throwing the
>>> error. I can't give out my VSS db, as that is intellectual
>>> property, but I would be willing to help track down this problem, if
>>> someone is willing to guide me a little.
>>
>>
>>
>> In order to help you we need to know a lot more about your setup. So
>> here it
>> goes:
>>
>> - Which OS are you running the conversion on? (Windows presumably)
>
Yes, windows. It is a version of the vss2svn.pl script floating around,
though I've made many changes to fix errors and adapt to our db. I am
running it via Cygwin (latest version of cygwin) and Perl for windows
(installed separate from cygwin).

>> - Which OS do you store your repository on?
>
Linux - RedHat 9

>> - Which access method does the conversion script use? (http, svn, file,
>> 'svnadmin load', ...)
>
http - Apache 2 using the svn client on the windows side (svn commands
such as commit, add, propset, etc)

>> - Where is your repository stored? - Is that an NFS or other
>> network-type drive?
>
No, it is a native filesystem on the linux side, I believe it is using
ext3, but it is possible that it is on ext2 (darn RedHat, doesn't use
something better such as Reiser, and I've yet to convert the drive
manually).

>>
>> Well, that should give us a first impression.
>>
>> bye,
>>
>>
>> Erik.
>>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Tue Jun 1 05:59:20 2004

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

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.