Re: site's been down
"James Henstridge" <[email protected]>
| Newsgroups | gmane.comp.python.db.psycopg.devel |
|---|---|
| Message-ID | <[email protected]> |
On 17/03/2008, Federico Di Gregorio <fog-NGVKUo/i/[email protected]> wrote: > On Sun, Mar 16, 2008 at 09:01:05PM -0600, Nathan wrote: > > I don't even get a 500 error. Just a "problem loading page" error on firefox. > > > > On Fri, Mar 14, 2008 at 6:09 PM, Michael Bayer <[email protected]> wrote: > > > http://www.initd.org/tracker/psycopg/wiki/PsycopgTwo keeps reporting > > > a 500 error for a couple of days now. Any word on the tracker being > > > up again ? > > > The problem was mod_python, sucking up all memory. So we switched to > FCGI and now a single trac process consumes about 4G of virtual memory > in 24h (then stops working). I am sorry but trac is such a piece of > crap that I heartily regret having started to use it. :( My guess is that this would be memory leaks in the Subversion bindings for Python. I know that the bzr-svn folks found many leaks, often requiring development releases to run without going out of memory. I wouldn't be surprised if the same holds for Trac. Perhaps setting a resource limit on the Trac fastcgi process would be a decent workaround (assuming that Apache is smart enough to restart dead fastcgi backends). James.