Hi Allen,
I verified that the rpm install has the patch, so you don't need to do
anything.
I created this patch about two weeks after you reported the problem.
When you start to run out of file descriptors, the error handling wasn't
robust, and you would leak some file descriptors. This made the problem
progressively worse.
You could reduce the number of parallel processes down to 2, but it
looks like I will have to go back to testing.
Could you check and make sure you aren't seeing any defunct processes or
leaked file descriptors? You can get a count of the file descriptors
like this:
$ ls -l /proc/`pgrep -f zencommand`/fd | wc -l
If this number grows over time, let me know.
-Eric
Allen Sanabria wrote:
running CentOS 4.4
1200 cmds every 5 mins, soon maybe more
and I installed from rpm.
thanks for the update I will try this.
So since I installed from RPM, that means I would have to download the
source and run this correct? Would this affect future RPM updates?
Meaning when the next update comes along and I run rpm -Uvh zenos.etc
would I have to download the source again and patch again?
So this was a fix as of recent cause, you worked on this issue with me
back then and this fix wasn't available?
On Tue, 2007-02-06 at 10:41 -0500, Eric Newton wrote:
Hi Allen,
How many processes do you run, and how often do they run? Also, what
platform are you using? I ran 1000 commands every 5 minutes during my
testing, and it didn't fall down.
There was a patch made to a library we used to run all these things in
parallel. I'm not sure it would have been applied on an update from
source. If you install using source (tarball, svn, etc), try this:
$ cd <wherever your zenoss inst directory is>
$ rm -rf build/Twisted-*
$ make twisted-install
That should patch and re-install Twisted.
-Eric
Allen Sanabria wrote:
Yea this happened to me last year and no matter what we tried it didn't
work.
I finally came up with a quick and dirty solution.
http://www.mail-archive.com/[email protected]/msg02455.html
check that out.
worked for me afterwards.
Error To many Processess!
Hey just letting you guys no the parallel 5 did in the zencommand.conf
did not work after I continued to add more checks.
So i broke it down by creating new monitors, when I did that I didn't
receive any more errors but that was because it was still only loading
the default monitor "localhost".
So after some thinking I came up with this solution.
I literally copied zencommand.conf to the new monitor names etc..
zencommand_Web.conf and put the variable monitor Web in that file.
Then I copied /bin/zencommand to /bin/zencommand_Web then in the new
file made sure the files match
contents of zencommand
#! /usr/bin/env bash
. $ZENHOME/bin/zenfunctions
PRGHOME=$ZENHOME/Products/ZenRRD
PRGNAME=zencommand.py
CFGFILE=$CFGDIR/zencommand.conf
PIDFILE=$VARDIR/$PRGNAME.pid
generic "$@"
contents of zencommand_Web
#! /usr/bin/env bash
. $ZENHOME/bin/zenfunctions
PRGHOME=$ZENHOME/Products/ZenRRD
PRGNAME=zencommand_Web.py
CFGFILE=$CFGDIR/zencommand_Web.conf
PIDFILE=$VARDIR/$PRGNAME.pid
generic "$@"
then for /Products/ZenRRD/zencommand.py i just mad another copy if it
and renamed it to zencommand_Web.py
then in /bin/zenoss I added this
C="$C zencommand_Web"
right under this
C="$C zencommand"
I also did the same for zenperfsnmp and zenperfxmlrpc
Now I can honestly say that I'm no longer having these issues.
Now I do not know if you guys had a easier or better solution but this
is what I came up with..
I tried adding multiple monitors and that didn't work and a few other
things that would not work at all.
___________________________________________
_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users
