[Catalyst] Session unexpectedly expiring

[Wade.Stuart at fallon.com]

Jeff Chimene <jchimene at gmail.com> wrote on 05/20/2007 10:13:33 PM:

> Thomas L. Shinnick wrote:
> > At 08:20 PM 5/20/2007, Jeff Chimene wrote:
> >> Hi,
> >> I'm trying to figure out why my sessions are expiring in
> >> script/xxx_cgi.pl but not script/xxx_server.pl
> > [snip]
> >> I'm using the Session, Session::Store::FastMmap,
> >> Session::State::Cookie plugins.
> >
> > I've not used it, I'm guessing, and I'm an idiot, but isn't this
> > likely permissions problems, between running standalone vs. running
> > under Apache?  Any chance the backing storage file is not available to
> > Apache processes?  Peeking at the docs:
> >     storage
> >          Specifies the file to be used for the sharing of session
> > data. ...
> >         Note that the file will be created with mode 0640, which means
> > that
> >          it will only be writeable by processes running with the same
> > uid as
> >          the process that creates the file.  ....
> >
> > Try deleting the backing file before re-running under Apache?
> Good catch. Apache (or script/aic_server.pl) refuses to honor a request
> if the working storage file created by the other HTTP server exists. I
> couldn't even get started w/o a preliminary "sudo rm -r /tmp/aic" So,
> although that's /a/ problem, it's not /the/ problem.

Without more info (you will post the info Matt asked for?)  this still
seems like the best candidate for the issue.  If you are running on Linux
and have selinux turned on, another common issue is that your policy does
not allow promiscuous writes from apache.  You may verify this is or is not
the issue by running audit2why < /var/log/audit/audit.log (or wherever your
se auditlog is located) and noting any blocks that are happening via your
web server process.

Also you may want to force the directory for the session stuffs to a sane
location ( don't know if you are or not -- you will post the info that Matt
asked for?).

-Wade