I'll set the zencommand in debug mode and see how it goes.  It's definitely
not ssh related.  The two commands that are running both run on the local
machine.  One of them is check_http to check a remote webpage and the other
is a custom Nagios check command that actually does a remote snmpget (it
needs to massage the result before returning it to Zenoss).

Both checks return nearly instantly when run from the command-line.

-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Eric Newton
Sent: Sunday, March 25, 2007 8:32 PM
To: Erik A. Dahl
Cc: General discussion of using zenoss system
Subject: Re: [zenoss-users] Big trouble with zencommand

Erik,

You may be thinking of a recent problem discovered in zenprocess.

    http://dev.zenoss.org/trac/ticket/1068

Todd,

Zenoss 1.1.1 already has the latest fix for working around a library bug 
affecting the scalability of zencommand. You will want to make sure you 
have patch01 for Twisted installed.  If you are using the Zenoss RPMs, 
or the appliance, this is done for you.  If you are using SVN or a 
source tarball you may have to clean zenossinst/build for this to be 
installed.

That said, there are no known problems where zencommand has an 
unnecessary time out.  Please switch your server into debug mode, and 
collect a log:

    $ mv $ZENHOME/zencommand.log $ZENHOME/zencommand.log.bak
    $ zencommand restart -v 10

Let zencommand run until you see the problem and send that log to us off 
list (compressed).

Here's a possible explanation for the problem: ZenCommand will keep as 
many as 256 SSH connections to remote hosts.  Normally this is useful 
because the setup for an SSH connection is quite high compared to the 
execution of a little nagios pluggin over that connection.  We have seen 
intermediary firewalls not appreciate Zenoss' efforts at maintaining a 
quiet SSH connection for 5 minutes.  If you are confident that your 
plugins are not SSH related, this is not your problem.

If you see any other related issues: memory usage or file-descriptor 
limits that occur at the same time, that would also help.

Thanks!

-Eric


Erik A. Dahl wrote:
> Eric,
>
> Have we fixed this in the trunk??  I know its come up.
>
> -EAD
>
> On Mar 24, 2007, at 12:05 PM, Todd Davis wrote:
>
>> This might be related to some problems I've been reporting (and haven't
>> heard anything about).
>>
>> I only have about 110 devices and only about 3 Nagios-type commands 
>> running
>> (and some ssh commands).  The issue I keep running into is the Nagios
>> commands time out quite often even though running them from the 
>> command-line
>> returns almost immediately.  After timing out, they usually clear 
>> within 5
>> minutes.
>>
>> -- 
>> Todd
>>
>> -----Original Message-----
>> From: [EMAIL PROTECTED]
>> [mailto:[EMAIL PROTECTED] On Behalf Of Allen Sanabria
>> Sent: Friday, March 23, 2007 2:55 PM
>> To: General discussion of using zenoss system
>> Subject: Re: [zenoss-users] Big trouble with zencommand
>>
>> Hey David, I had this same issues last year, and how I solved it, was by
>> breaking down those 800 or so devices into multiple monitors.
>>
>> and by doing that you will have to create new zencommand's
>> I created a how to a month or so back and someone else created a similar
>> how to a few days ago.
>>
>> You are essentially creating another zencommand and putting it into
>> another monitor.
>>
>> Reason for all this is to break down how many open files are running
>> simultaneously under on each monitor
>>
>> here is essentially what I have been doing
>>
>> Step 1
>>
>> create new performance monitor
>>
>> Step 2
>>
>> I copied over bin/zenperfsnmp to bin/zenperfsnmp_netapp
>> contents of zenpersnmp_netapp
>>
>> (I changed PRGNAME and CFGFILE)
>>
>> PRGHOME=$ZENHOME/Products/ZenRRD
>> PRGNAME=zenperfsnmp_netapp.py
>> CFGFILE=$CFGDIR/zenperfsnmp_netapp.conf
>> PIDFILE=$VARDIR/$PRGNAME.pid
>>
>>
>> Step 3
>>
>>
>> then for /Products/ZenRRD/zenperfsnmp.py i just mad another copy if it
>> and renamed it to zenperfsnmp_netapp.py and made all instance of
>> zenperfsnmp point to this instead zenperfsnmp_netapp.
>> This will create its own log file.
>>
>>
>> Step 4
>> then in /bin/zenoss I added zenperfsnmp_netapp under zenperfsnmp
>>
>> C="$C zenperfsnmp
>> C="$C zenperfsnmp_netapp"
>>
>> step 5
>>
>> copy over etc/zenperfsnmp.conf to etc/zenperfsnmp_netapp.conf
>>
>> monitor         NetApp
>>
>> Step 6
>>
>> restart zenoss
>>
>> Step 7 add NetApps to the NetApp performance monitor
>>
>>
>> _______________________________________________
>> zenoss-users mailing list
>> [email protected]
>> http://lists.zenoss.org/mailman/listinfo/zenoss-users
>>
>> _______________________________________________
>> zenoss-users mailing list
>> [email protected]
>> http://lists.zenoss.org/mailman/listinfo/zenoss-users
>
>

_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users

_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users

Reply via email to