Could people *please* start mentioning bdb versions when reporting this kind
of wedging! There are 3 different versions of bdb with different track
records for stability. You don't do the community any good by reporting this
kind of general statement. Even better would be if you reported:
- bdb version;
- system architecture;
- OS version;
- Single / multiple processor system;
- Storage type (IDE/SCSI HD) for repository;
- Filesystem type for repository;
Any other information which might be relevant to the way bdb functions.
bye,
Erik.
PS: Alan, this is not personal. We have extremely many of these reports, but
they do not help building stats or solving the problems because they are
just too vague.
>
> I had repeated wedging over a year, installed a brand new server, all
> with debian packages (from what I remember), and the wedging continued
> (after a week or so) on the fresh install.
>
> The only possible reasons for the wedging that I could think of.
> - issue with prebuilt packages (apr/bdb/subversion etc.) being slightly
> incompatible, maybe they where compiled with different gcc/version/flags
> etc.
> - poor fault recovery in bdb? - If something goes foobar in fsfs, it is
> usually a very simple filesystem call failure, (where as there are
> potentially mulitple points of failure in bdb)
> - disk getting full (on old server) - causing all sorts of havoc..
>
> It may be that some minor issue in the setup I have caused bdb wedging
> (like the copy on commit scripts), but the design of fsfs, appears to be
> alot more fault tolerant the bdb.
>
>
> Regards
> Alan
>
>
> Patrick Dean Rusk wrote:
>
> >"Martin Tomes" <lists@tomes.org> wrote...
> >
> >
> >>We use BDB and it has *never* got wedged and there is little reason why
> it
> >>
> >>
> >should if you set up the
> >
> >
> >>server correctly.
> >>
> >>
> >
> >That has been our experience as well.
> >
> >It would be helpful if people like Scott who talk Subversion wedging
> their
> >repositories a lot in the past would make it clear what version of
> >Subversion they were using. My impression is that there is a lot less
> >wedging going on now than before 1.0. We're up to 1200 revisions without
> >having had a wedge or a single dangling transaction (the kind you see
> with
> >"svnadmin lstxns").
> >
> >On the other hand, I'm not surprised at problems arising when people
> convert
> >large VSS or CVS repositories to SVN. It's always struck me as a fool's
> >errand to do so, but I guess it's important to some people to not have to
> go
> >to another tool to see history on a file.
> >
> >Patrick Rusk
> >
> >
> >
> >
> >---------------------------------------------------------------------
> >To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> >For additional commands, e-mail: users-help@subversion.tigris.org
> >
> >
> >
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
>
--
NEU: GMX ProMail mit bestem Virenschutz http://www.gmx.net/de/go/mail
+++ Empfehlung der Redaktion +++ Internet Professionell 10/04 +++
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Sat Sep 18 13:28:24 2004