Hello List,
I’ve been noticing that my RT web server has been running out of memory
very rapidly lately. And when I restart the web server I see this strange
RT related message:
[monitor03:~] root% apachectl -t
Odd number of elements in anonymous hash at
/usr/local/rt4/sbin/…/lib/RT/Config.pm line 1055.
Odd number of elements in hash assignment at
/usr/local/rt4/sbin/…/lib/RT/Config.pm line 1056.
Syntax OK
And I was wondering if this message could be related to the unusual level
of memory that Apache is gobbling up. I know that I have restarted the web
server several times in the past few months and never seen this message.
This is new and so is the phenomenon of the web server quickly running out
of memory. Which is why I surmise they may be related.
I know that I have never personally altered the file. It seems to have been
unchanged since the initial RT install in Oct of 2011:
48 -rw-r–r-- 1 root bin 45585 Oct 13 2011 /usr/local/rt4/lib/RT/Config.pm
I am not the only one who admins RT in this 24 web shop. But I have made
some minor tweaks to RT_Site_Config.pm . Most recently I enabled GPG
settings there. Could any change to RT_Site_Config.pm cause this type of
message? Or alter RT so that it consumes copious amounts of memory? I
notice also that when I STOP the web server for a moment or two, I
instantly have two more gigabytes of memory available.
But then I turn the web server back on and its fine for a few minutes. But
then RT’s performance slows to a crawl and I notice that the web server
only has 10 MB free.
This is what I see in the RT apache error log:
monitor03:~] root% tail -f /usr/local/rt4/var/log/apache2.error
[Sat Jan 26 07:07:31 2013] [error] [client 10.40.27.251]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153
[Sat Jan 26 07:17:18 2013] [error] [client 10.40.27.251]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153
[Sat Jan 26 07:19:56 2013] [error] [client 10.75.13.45] Directory index
forbidden by Options directive: /usr/local/rt4/share/html
[Sat Jan 26 07:19:56 2013] [error] [client 10.75.13.45] File does not
exist: /usr/local/rt4/share/htmlerror
[Sat Jan 26 07:19:56 2013] [error] [client 10.75.13.45] File does not
exist: /usr/local/rt4/share/htmlfavicon.ico
[Sat Jan 26 07:19:56 2013] [error] [client 10.75.13.45] File does not
exist: /usr/local/rt4/share/htmlfavicon.ico
[Sat Jan 26 10:34:13 2013] [error] [client 10.75.13.45]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153, referer:
http://rt.mm3.mlbam.com/helpdesk/Search/Simple.html?q=Muthuraj
[Sat Jan 26 10:34:29 2013] [error] [client 10.75.13.45]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153, referer:
http://rt.mm3.mlbam.com/helpdesk/Search/Simple.html?q=Muthuraj
[Sat Jan 26 10:36:03 2013] [error] [client 10.75.13.45]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153, referer:
http://rt.mm3.mlbam.com/helpdesk/Search/Simple.html?q=Muthuraj
[Sat Jan 26 10:45:29 2013] [error] [client 10.40.27.251]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153
Top confirms that apache is the number 1 consumer of memory. It’s a strange
situation and I’d appreciate any advice you might have on how to proceed
from here.
Thanks
Tim
GPG me!!
gpg --keyserver pool.sks-keyservers.net --recv-keys F186197B
Also this server has 2GB of ram. Generally speaking is that enough to run
RT 4.02 on?
Thanks
timOn Sat, Jan 26, 2013 at 11:19 AM, Tim Dunphy bluethundr@gmail.com wrote:
Hello List,
I’ve been noticing that my RT web server has been running out of memory
very rapidly lately. And when I restart the web server I see this strange
RT related message:
[monitor03:~] root% apachectl -t
Odd number of elements in anonymous hash at
/usr/local/rt4/sbin/…/lib/RT/Config.pm line 1055.
Odd number of elements in hash assignment at
/usr/local/rt4/sbin/…/lib/RT/Config.pm line 1056.
Syntax OK
And I was wondering if this message could be related to the unusual level
of memory that Apache is gobbling up. I know that I have restarted the web
server several times in the past few months and never seen this message.
This is new and so is the phenomenon of the web server quickly running out
of memory. Which is why I surmise they may be related.
I know that I have never personally altered the file. It seems to have
been unchanged since the initial RT install in Oct of 2011:
48 -rw-r–r-- 1 root bin 45585 Oct 13 2011
/usr/local/rt4/lib/RT/Config.pm
I am not the only one who admins RT in this 24 web shop. But I have made
some minor tweaks to RT_Site_Config.pm . Most recently I enabled GPG
settings there. Could any change to RT_Site_Config.pm cause this type of
message? Or alter RT so that it consumes copious amounts of memory? I
notice also that when I STOP the web server for a moment or two, I
instantly have two more gigabytes of memory available.
But then I turn the web server back on and its fine for a few minutes. But
then RT’s performance slows to a crawl and I notice that the web server
only has 10 MB free.
This is what I see in the RT apache error log:
monitor03:~] root% tail -f /usr/local/rt4/var/log/apache2.error
[Sat Jan 26 07:07:31 2013] [error] [client 10.40.27.251]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153
[Sat Jan 26 07:17:18 2013] [error] [client 10.40.27.251]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153
[Sat Jan 26 07:19:56 2013] [error] [client 10.75.13.45] Directory index
forbidden by Options directive: /usr/local/rt4/share/html
[Sat Jan 26 07:19:56 2013] [error] [client 10.75.13.45] File does not
exist: /usr/local/rt4/share/htmlerror
[Sat Jan 26 07:19:56 2013] [error] [client 10.75.13.45] File does not
exist: /usr/local/rt4/share/htmlfavicon.ico
[Sat Jan 26 07:19:56 2013] [error] [client 10.75.13.45] File does not
exist: /usr/local/rt4/share/htmlfavicon.ico
[Sat Jan 26 10:34:13 2013] [error] [client 10.75.13.45]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153, referer:
http://rt.mm3.mlbam.com/helpdesk/Search/Simple.html?q=Muthuraj
[Sat Jan 26 10:34:29 2013] [error] [client 10.75.13.45]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153, referer:
http://rt.mm3.mlbam.com/helpdesk/Search/Simple.html?q=Muthuraj
[Sat Jan 26 10:36:03 2013] [error] [client 10.75.13.45]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153, referer:
http://rt.mm3.mlbam.com/helpdesk/Search/Simple.html?q=Muthuraj
[Sat Jan 26 10:45:29 2013] [error] [client 10.40.27.251]
Apache2::RequestIO::rflush: (103) Software caused connection abort at
/usr/lib/perl5/site_perl/5.8.8/Plack/Handler/Apache2.pm line 153
Top confirms that apache is the number 1 consumer of memory. It’s a
strange situation and I’d appreciate any advice you might have on how to
proceed from here.
Thanks
Tim
–
GPG me!!
gpg --keyserver pool.sks-keyservers.net --recv-keys F186197B
GPG me!!
gpg --keyserver pool.sks-keyservers.net --recv-keys F186197B
Also this server has 2GB of ram. Generally speaking is that enough to run
RT 4.02 on?
Thanks
tim
Hi Tim,
If the DB is on the same machine, not really. Version 3.2 worked okay with
2GB, 3.6 worked okay with 4-8GB and that is the least amount I would try
to use for 4.0 with them on the same machine.
Cheers,
Ken
I’ve been noticing that my RT web server has been running out of memory very rapidly lately.
And when I restart the web server I see this strange RT related message:
[monitor03:~] root% apachectl -t
Odd number of elements in anonymous hash at /usr/local/rt4/sbin/…/lib/RT/Config.pm line 1055.
Odd number of elements in hash assignment at /usr/local/rt4/sbin/…/lib/RT/Config.pm line
1056.
Syntax OK
This implies that something in RT_SiteConfig.pm is specified
incorrectly. You say you set up GnuPG which uses hashes. I’d check
that setup carefully, or show it to the list.
But then I turn the web server back on and its fine for a few minutes. But then RT’s
performance slows to a crawl and I notice that the web server only has 10 MB free.
This is what I see in the RT apache error log:
To diagnose something like this you need to say a lot more about your
setup. Is this the webserver and dbserver or only the webserver?
What deployment option are you using (mod_perl, fcgi, something else)?
How large is each individual apache child, is there one huge one or
are they all large? Assuming a webserver only with apache, you may
need to tell Apache to only start a few children, and limit the max
children so that it can’t grow unbounded.
-kevin