Since putting my RT2 install live, i’ve noticed a savage memory leak
appearing in apache/mod_perl, my apache processes are getting up to 200mb+
in size. I’m using
apache 1.3.22
mod_perl 1.26 (static)
RT2 2.0.8
Anyone else noticed this problem? I should probably also point out that I
think my rt2 database is somewhat larger than most peoples here (at 250,000
tickets). I’d really like to talk to anyone with exprience running rt2
with such a large ticket base.
Matthew Watson
Development, Netspace Online Systems
mwatson@netspace.net.au
Since putting my RT2 install live, i’ve noticed a savage memory leak
appearing in apache/mod_perl, my apache processes are getting up to 200mb+
in size. I’m using
apache 1.3.22
mod_perl 1.26 (static)
RT2 2.0.8
Anyone else noticed this problem? I should probably also point out that I
think my rt2 database is somewhat larger than most peoples here (at 250,000
tickets). I’d really like to talk to anyone with exprience running rt2
with such a large ticket base.
FastCGI might be the answer …
I’ve been very curious to hear about people using FastCGI anyway. Anybody
have any mod_perl vs. FastCGI comparisons?
Since putting my RT2 install live, i’ve noticed a savage memory leak
appearing in apache/mod_perl, my apache processes are getting up to 200mb+
in size. I’m using
apache 1.3.22
mod_perl 1.26 (static)
RT2 2.0.8
Anyone else noticed this problem? I should probably also point out that I
think my rt2 database is somewhat larger than most peoples here (at 250,000
tickets). I’d really like to talk to anyone with exprience running rt2
with such a large ticket base.
I’ve run test databases double that size without seeing any sort of leakage
like that. How easily reproducable is that memory leakage? How long does
it take to get that big? Do processes just constantly grow, or do
they grow and shrink? One quick fix is to configure your apache to reap
processes after a smaller number of connections. (How do you have it tuned
now?)
We moved from mod_perl to FastCGI with much success.
FastCGI appears to run RT noticably faster than mod_perl.
We also have less apache seg faults with FastCGI.
Overall FastCGI has been a good move for us.
Sytem is Solaris 8.On Mon, 7 Jan 2002, Charlie Watts wrote:
On Tue, 8 Jan 2002, Matthew Watson wrote:
Since putting my RT2 install live, i’ve noticed a savage memory leak
appearing in apache/mod_perl, my apache processes are getting up to 200mb+
in size. I’m using
apache 1.3.22
mod_perl 1.26 (static)
RT2 2.0.8
Anyone else noticed this problem? I should probably also point out that I
think my rt2 database is somewhat larger than most peoples here (at 250,000
tickets). I’d really like to talk to anyone with exprience running rt2
with such a large ticket base.
FastCGI might be the answer …
I’ve been very curious to hear about people using FastCGI anyway. Anybody
have any mod_perl vs. FastCGI comparisons?
Its very reproducable, It happens at the exact same time as the possible bug
I mentioned in devel. eg, when a user changes the queue for a ticket, from a
queue he can see (obviously) to a queue he cannot see… The browser just
hangs and hangs (opera and IE6), and the httpd process for that user sky
rockets from about 20mb to ~250 mb (where it suddenly starts hitting
swapspace).
I’ve currently got apache set reasonably strict as far as connections go,
the main issue comes when uses think “oh this isn’t working, i’ll press
refresh” , which just starts up another big ugly httpd.