Ad Widget

**caarts** · 08-05-2009, 14:07

Not sure (correct me if I am mistaken),
A 0 indicates that either the process does not exist (or that it cannot find it), or mispelled name of the process you wish to monitor, or that it really is not running at all.

I have several SuSE 9.1/9.3/10 servers at work that are monitored by Zabbix.

Running #zabbix_agentd -t proc.num[apache2] will output

#proc.num[apache2] [u|0]

But, running #zabbix_agentd -t proc.num[httpd2-prefork] -> will result in:

#proc.num[httpd2-prefork] [u|11]

Do a ps -ef | grep (process you are looking for) and check the output.

**wbro1** · 11-05-2009, 08:16

Thanks for the response, I tried it right away, see results below.
What can I do with this info.

CN01:~ # ps -ef |grep novell-ncs
root 25058 23890 0 08:09 pts/0 00:00:00 grep novell-ncs
CN01:~ # ps -ef |grep novell-nss
root 26828 23890 0 08:10 pts/0 00:00:00 grep novell-nss
CN01:~ # ps -ef |grep ntp
ntp 4386 1 0 Apr03 ? 00:00:00 /usr/sbin/ntpd -p /var/lib/ntp/var/run/ntp/ntpd.pid -u ntp -i /var/lib/ntp
root 27443 23890 0 08:10 pts/0 00:00:00 grep ntp
CN01:~ # ps -ef |grep apache2
root 6421 1 0 Apr15 ? 00:00:00 /usr/sbin/httpd2-worker -f /etc/apache2/httpd.conf -DSSL
wwwrun 12065 6421 0 04:12 ? 00:00:00 /usr/sbin/httpd2-worker -f /etc/apache2/httpd.conf -DSSL
wwwrun 13462 6421 0 Apr23 ? 00:00:00 /usr/sbin/httpd2-worker -f /etc/apache2/httpd.conf -DSSL
root 29811 23890 0 08:12 pts/0 00:00:00 grep apache2

**caarts** · 11-05-2009, 10:09

wbro 1

When you: # ps -ef |grep novell-ncs and it returns: root 25058 23890 0 08:09 pts/0 00:00:00 grep novell-ncs -> it means there is no such process, either other name or it is not running.

**wbro1** · 11-05-2009, 10:17

I can see it running with the status command below.
How can I check this in Zabbix ? Isn't this a process ?

CN01:~ # rcnovell-ncs status
running
CN01:~ #

for your info, command below gives same result.

CN01:/etc/rc.d # ./novell-ncs status
running
CN01:/etc/rc.d #

**richlv** · 11-05-2009, 20:13

you have to find out actual process name, not the name of the initscript.

for a grepped list, i'd suggest using "ps -AH|grep <string>".
for a definitive match on processname, use "ps -C <processname>"

**wbro1** · 12-05-2009, 15:59

Here's what I did so far,

Looked in novell-ncs with cat novell-ncs.
Found near the start parameter the calling of ldncs

Looked in ldncs with cat ldncs.
Found /sbin/evmsd

this one I can find in ps -AH |grep evmsd

What i don't know, is this the only thing novell-ncs is running.
How can I find out values below with Zabbix

chkconfig --list novell-ncs
novell-ncs 0

ff 1

ff 2

ff 3

n 4

ff 5

n 6

ff

**caarts** · 12-05-2009, 16:21

Try the ps suggestions richlv gave and paste the output here in a new reply.

**richlv** · 12-05-2009, 16:29

first, yes, initscript could be starting multiple processes.
second, values you posted would make little sense to monitor - you would only find out whether service is _configured_ to run, not whether it is running. while a valid goal, i understood that you want to check whether the service is actually running.

you could rely on initscripts to work correctly and use service status, though. for that you could use user parameters (read in the manual about them) by configuring userparameter with a string, similar to this :

Code:

service <service> status > /dev/null 2>&1; echo $?

and checking whether returned values is 0. if it is, service is reported as running. if it's some other value, service is not running.

for bonus points use positional parameter instead of a fixed service name in the userparameter.

**rue** · 12-05-2009, 16:40

If evmsd is the only process started directly in the init script, try:

Code:

ps -ef --forest

...to see what's running what on your system.

cheers

Rü

**richlv** · 12-05-2009, 16:49

it would be helpful if you could find out process names yourself.
as a quick way you could try something like this to find out what processes the service might use...

Code:

SERVICE=novell-ncs; service $SERVICE stop; ps -Ao comm= | sort -u|less > no; service $SERVICE start; ps -Ao comm= | sort -u|less > yes; diff yes no

note, it restarts a particular service once. it will most likely also catch other processes, but you should be able to decypher which ones do not belong to a particular service

**wbro1** · 15-05-2009, 12:16

@ richlv

I think your option is the best for us.

I edited the zabbix_agentd.conf and added your line without the > /dev/null 2>&1; echo $?

UserParameter=OESstatus[*] $1 status

restarted with zabbix_agentd restart.

Then added an item with key

OESstatus[rcnovell-ncs]

resulting in value 0

0 is running, other values is not running I presume ?

Thanks for your input.

**wbro1** · 15-05-2009, 12:34

This only works with 'text' values i guess. I get following message on another item.

Type of received value [NSS kernel modules..running NSS admin volume..unused] is not sutable for [OESstatus[rcnovell-nss]@CN03]

the console output of rcnovell-nss status is:

NSS kernel modules running
NSS admin volume running

**richlv** · 15-05-2009, 13:28

and why did you remove "> /dev/null 2>&1; echo $?" part ?

**wbro1** · 15-05-2009, 13:55

To be sure I knew what the sentence did.
Your parameters I can't translate to what it wil do

Ad Widget

Number of running processes = 0

Number of running processes = 0

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment