Using Oracle DB for Apache::Session data

I’ve been trying to figure out why my production instance of RT running
on Oracle is slower than my Dev instance on MySQL. I saw something on
the mailing list that I thought could offer some help - someone
mentioned cleaning out the sessions table. Okay, I’ll take a look at
that, I thought. But then I was surprised to find the table empty!
Instead, the session information is being written to disk at
/path-to-rt/var/session_data/. Ah-ha! I found that the file
Elements/SetupSessionCookie had this line:

Oracle => ‘Apache::Session::Oracle’,

And of course, it’s commented out, when the mysql and Pg lines were not.
So I uncommented this, restarted apache; the home page loaded about 20%
faster, and a row showed up in the Oracle database (dev now pointing to
a QA Oracle backend rather than MySQL for this test). Sweet! But then
I clicked on the logout button, and the page just hangs. Anyone have
any thoughts? I’m poking around at it now.

Eric Schultz

Call on God, but row away from the rocks.

  • Indian proverb

Nevermind, my problem appears to be related to changing how I did my
sessions without having first logged out :slight_smile: Subsequent sessions worked
fine. However, it’s still interesting to note that this feature is
disabled in the latest versions of 3.4.x and 3.5.x.

I’ve been trying to figure out why my production instance of RT
running
on Oracle is slower than my Dev instance on MySQL. I saw something on
the mailing list that I thought could offer some help - someone
mentioned cleaning out the sessions table. Okay, I’ll take a look at
that, I thought. But then I was surprised to find the table empty!
Instead, the session information is being written to disk at
/path-to-rt/var/session_data/. Ah-ha! I found that the file
Elements/SetupSessionCookie had this line:

Oracle => ‘Apache::Session::Oracle’,

And of course, it’s commented out, when the mysql and Pg lines were
not.
So I uncommented this, restarted apache; the home page loaded about
20%
faster, and a row showed up in the Oracle database (dev now
pointing to
a QA Oracle backend rather than MySQL for this test). Sweet! But
then
I clicked on the logout button, and the page just hangs. Anyone have
any thoughts? I’m poking around at it now.

I tried this, in hopes of resolving a persistent logout issue we have
where every link you click on you are asked to log back in. When I
uncommented the line and restarted apache (after removing the
contents of my session cache directory) it hangs when logging in and
never actually loads the rt page. There is one entry in the session
table though, so it gets at least that far.

My RT log shows this:
[Wed Feb 1 20:39:49 2006] [info]: Successful login for plummer from
0.0.0.0 (/opt/rt3/share/ht
ml/autohandler:215)
[Wed Feb 1 20:39:53 2006] [debug]: RT::Date used date::parse to make
1970-01-01 18000 (/opt/rt3/lib
/RT/Date.pm:212)

Every login seems to show this date error, could that be the source
of my forcing a login on each link clicked issue?

Shawn Plummer
Systems Manager
CIT SUNY Geneseo
"The mind can make substance, and people planets of its own with
beings brighter than have been, and give a breath to forms which can
outlive all flesh." -Lord Byron

smime.p7s (2.42 KB)

Nevermind, my problem appears to be related to changing how I did my
sessions without having first logged out :slight_smile: Subsequent sessions worked
fine. However, it’s still interesting to note that this feature is
disabled in the latest versions of 3.4.x and 3.5.x.

It’s disabled because it at least used to be horribly, horribly broken.

Interesting. I didn’t have the persistent session problem you describe.
What version of RT are you running? If seems like it would be worth
investigating that other log statement you get, especially if it happens
both when you try to do session maintenance on disk as well as in the
DB.