Simon Perreault <nomis80@nomis80.org> writes:
> Trying to do some basic timing measurements, I wrote the following Python
> code:
>
> ---------------------------------------------------------------------------
> ...
> def main(pool, url):
> revision = svn.core.svn_opt_revision_t()
> revision.kind = svn.core.svn_opt_revision_head
> config = svn.core.svn_config_get_config( None, pool )
> ctx = svn.client.svn_client_create_context(pool)
> ctx.auth_baton = svn.core.svn_auth_open([], pool)
> ctx.config = config
> for i in range(10000):
> print i
> files = svn.client.svn_client_ls( url, revision, 0, ctx, pool )
> ...
> ---------------------------------------------------------------------------
>
> Running this program always gives the following output:
>
> ---------------------------------------------------------------------------
> ...
> 1015
> 1016
> 1017
> 1018
> Traceback (most recent call last):
> File "./svnls", line 34, in ?
> svn.core.run_app(main, sys.argv[1])
> File "/usr/lib/python2.3/site-packages/svn/core.py", line 33, in run_app
> return apply(func, (pool,) + args, kw)
> File "./svnls", line 16, in main
> files = svn.client.svn_client_ls( url, revision, 0, ctx, pool )
> libsvn._core.SubversionException: ('Unable to open an ra_local session to
> URL', 180001)
> ---------------------------------------------------------------------------
>
> Why does it fail at the 1020th call? Is it an anti-hacker feature,
> is my code wrong, or is it *gasp* a bug? Where does that magic 1020
> number come from?
You're probably running into a Berkeley DB lock limit or something.
Does this reproduce with an FSFS repository? If you up some lock
parameters in repos/db/DB_CONFIG (and then run recovery on the
repository), does it change the behavior?
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Thu Feb 17 06:07:46 2005