Apache & zabbix

[steveboyson]

Hi folks,
I made a different version of https://www.zabbix.com/wiki/templates/apache
My version works with zabbix_sender and trapper items.
It caches the output and fetches only if the cached file is too old (can be set via parameter INTERVAL).


PROS: less work & load on zabbix agent side
CONS: values may be less accurate due to caching

Download script and template here:

https://www.netzwerk-aktiv.com/owncloud/public.php?service=files&t=078d705a4e13ae1dc46627a5e1eeed55

[sccuser]

Dear stevenboyson,

Thanks for the great script and template.
I did exactly as your instruction but no luck to display the graphs.

Help me pls.

[steveboyson]

What happens if you execute the script from within your zabbix server as
"bash -x apache_zabbix.sh"?

Have you checked the variables in the script? Are the settings

zbx_send="/usr/bin/zabbix_sender -c /etc/zabbix/zabbix_agentd.conf"
zbx_server="zabbix"

correct?
Is wget in the correct location or correct configured in the script?
Is apache's mod_status enabled? Can you access it via

http://<your-server-name>/server-status?auto

?

Pls. check all that.

[sccuser]

What happens if you execute the script from within your zabbix server as
"bash -x apache_zabbix.sh"?

Have you checked the variables in the script? Are the settings

zbx_send="/usr/bin/zabbix_sender -c /etc/zabbix/zabbix_agentd.conf"
zbx_server="zabbix"

correct?
Is wget in the correct location or correct configured in the script?
Is apache's mod_status enabled? Can you access it via

http://<your-server-name>/server-status?auto

?

Pls. check all that.

Dear steveboyson,

Forgot to let you know that I'm trying to monitor apache from remote machine, not Zabbix server itself.

When trying to run bash -x /home/zabbix/bin/zabbix_apache.sh (I locate the script in here), it returns

[root@xentos ~]# bash -x /home/zabbix/bin/zabbix_apache.sh
+ zapachever=1.0_MH
+ rval=0
+ RES=
+ INTERVAL=3
+ zbx_send='/usr/local/zabbix_agentd/bin/zabbix_sender -c /usr/local/etc/zabbix_agentd.conf'
+ zbx_server=zabbix
++ hostname -s
+ zbx_host=xentos
+ prefix=apache
+ [[ 0 == 0 ]]
+ APACHE=localhost
+ apache_status_file=/tmp/zbx_apache_localhost.txt
++ find /tmp/zbx_apache_localhost.txt -cmin +3
+ '[' x/tmp/zbx_apache_localhost.txt '!=' x ']'
+ fetch
+ wget --quiet -O /tmp/zbx_apache_localhost.txt 'http://localhost/server-status?auto'
+ rval=0
++ do_zbx_send Stats_source FETCH
+++ /usr/local/zabbix_agentd/bin/zabbix_sender -c /usr/local/etc/zabbix_agentd.conf -z zabbix -s xentos -k 'apache[Stats_source]' -o FETCH
++ RES='Sending failed. Use option -vv for more detailed output.'
++ echo Sending failed. Use option -vv for more detailed output.
+ RES='Sending failed. Use option -vv for more detailed output.'
+ keys='TotalAccesses TotalKBytes Uptime ReqPerSec BytesPerSec BytesPerReq BusyWorkers IdleWorkers version WaitingForConnection StartingUp ReadingRequest SendingReply'
+ keys='TotalAccesses TotalKBytes Uptime ReqPerSec BytesPerSec BytesPerReq BusyWorkers IdleWorkers version WaitingForConnection StartingUp ReadingRequest SendingReply KeepAlive DNSLookup ClosingConnection Logging GracefullyFinishing IdleCleanupOfWorker OpenSlotWithNoCurrentProcess'
++ cat /tmp/zbx_apache_localhost.txt
+ VAR='Total Accesses: 1
Total kBytes: 544
CPULoad: .0588235
Uptime: 68
ReqPerSec: .0147059
BytesPerSec: 8192
BytesPerReq: 557056
BusyWorkers: 1
IdleWorkers: 10
Scoreboard: ______W____....................................... .................................................. .................................................. .................................................. .................................................. ......'
+ OUT=

and so on...

I also modify the script to appropriate

zapachever="1.0_MH"
rval=0
RES=''
# minutes
INTERVAL=3

zbx_send="/usr/local/zabbix_agentd/bin/zabbix_sender -c /usr/local/etc/zabbix_agentd.conf"
zbx_server="zabbix"
zbx_host=`hostname -s`
prefix=apache

do_zbx_send ()

Please tell me how to do the below

Is wget in the correct location or correct configured in the script?

Is apache's mod_status enabled? YES

[steveboyson]

+++ /usr/local/zabbix_agentd/bin/zabbix_sender -c /usr/local/etc/zabbix_agentd.conf -z zabbix -s xentos -k 'apache[Stats_source]' -o FETCH

and

++ RES='Sending failed. Use option -vv for more detailed output.'
++ echo Sending failed. Use option -vv for more detailed output.
+ RES='Sending failed. Use option -vv for more detailed output.'

need to be checked. Is "zabbix" the name of your zabbix server?

wget seems to be installed correctly, by the way.

Pls. give some more details about your configuration.

[sccuser]

Dear steveboyson,

My apologize about the incorrect zabbix server name in the script, I changed to proper one and run "bash -x ..." again then receive the below, it seem likely working with a little bit error this time (firewall?), but the graph cannot be displayed as proper.

[root@xentos ~]# bash -x /home/zabbix/bin/zabbix_apache.sh
+ zapachever=1.0_MH
+ rval=0
+ RES=
+ INTERVAL=3
+ zbx_send='/usr/local/zabbix_agentd/bin/zabbix_sender -c /usr/local/etc/zabbix_agentd.conf'
+ zbx_server=xibbaz
++ hostname -s
+ zbx_host=xentos
+ prefix=apache
+ [[ 0 == 0 ]]
+ APACHE=localhost
+ apache_status_file=/tmp/zbx_apache_localhost.txt
++ find /tmp/zbx_apache_localhost.txt -cmin +3
+ '[' x/tmp/zbx_apache_localhost.txt '!=' x ']'
+ fetch
+ /usr/bin/wget --quiet -O /tmp/zbx_apache_localhost.txt 'http://localhost/server-status?auto'
+ rval=0
++ do_zbx_send Stats_source FETCH
+++ /usr/local/zabbix_agentd/bin/zabbix_sender -c /usr/local/etc/zabbix_agentd.conf -z xibbaz -s xentos -k 'apache[Stats_source]' -o FETCH
++ RES='info from server: "processed: 0; failed: 1; total: 1; seconds spent: 0.000031"
sent: 1; skipped: 0; total: 1'
++ echo info from server: '"processed:' '0;' failed: '1;' total: '1;' seconds spent: '0.000031"' sent: '1;' skipped: '0;' total: 1
+ RES='info from server: "processed: 0; failed: 1; total: 1; seconds spent: 0.000031" sent: 1; skipped: 0; total: 1'
+ keys='TotalAccesses TotalKBytes Uptime ReqPerSec BytesPerSec BytesPerReq BusyWorkers IdleWorkers version WaitingForConnection StartingUp ReadingRequest SendingReply'
+ keys='TotalAccesses TotalKBytes Uptime ReqPerSec BytesPerSec BytesPerReq BusyWorkers IdleWorkers version WaitingForConnection StartingUp ReadingRequest SendingReply KeepAlive DNSLookup ClosingConnection Logging GracefullyFinishing IdleCleanupOfWorker OpenSlotWithNoCurrentProcess'
++ cat /tmp/zbx_apache_localhost.txt
+ VAR='Total Accesses: 10888
Total kBytes: 207350
CPULoad: .0585563
Uptime: 57671
ReqPerSec: .188795
BytesPerSec: 3681.68
BytesPerReq: 19501
BusyWorkers: 1
IdleWorkers: 16
Scoreboard: ____W____________................................. .................................................. .................................................. .................................................. .................................................. ......'
+ OUT=
+ for key in '$keys'
++ getValue TotalAccesses
++ CASE_VALUE=TotalAccesses
++ case $CASE_VALUE in
++ grep 'Total Accesses:'
++ echo 'Total Accesses: 10888
Total kBytes: 207350
CPULoad: .0585563
Uptime: 57671
ReqPerSec: .188795
BytesPerSec: 3681.68
BytesPerReq: 19501
BusyWorkers: 1
IdleWorkers: 16
Scoreboard: ____W____________................................. .................................................. .................................................. .................................................. .................................................. ......'
++ awk '{print $3}'
++ rval=0
+ val=10888
++ do_zbx_send TotalAccesses 10888
+++ /usr/local/zabbix_agentd/bin/zabbix_sender -c /usr/local/etc/zabbix_agentd.conf -z xibbaz -s xentos -k 'apache[TotalAccesses]' -o 10888
++ RES='info from server: "processed: 0; failed: 1; total: 1; seconds spent: 0.000031"
sent: 1; skipped: 0; total: 1'
++ echo info from server: '"processed:' '0;' failed: '1;' total: '1;' seconds spent: '0.000031"' sent: '1;' skipped: '0;' total: 1
+ RES='info from server: "processed: 0; failed: 1; total: 1; seconds spent: 0.000031" sent: 1; skipped: 0; total: 1'
+ for key in '$keys'
++ getValue TotalKBytes
++ CASE_VALUE=TotalKBytes
++ case $CASE_VALUE in
++ grep 'Total kBytes:'
++ echo 'Total Accesses: 10888
Total kBytes: 207350
CPULoad: .0585563
Uptime: 57671
ReqPerSec: .188795
BytesPerSec: 3681.68
BytesPerReq: 19501
BusyWorkers: 1
IdleWorkers: 16
Scoreboard: ____W____________................................. .................................................. .................................................. .................................................. .................................................. ......'
++ awk '{print $3}'
++ rval=0
+ val=207350

and so on...

Here below is what I did:
1. copy the zabbix_apache.sh to /home/zabbix/bin

2. chmod +x /home/zabbix/bin/zabbix_apache.sh + granting permission for "zabbix" group.



3. modify zabbix_apache.sh

zbx_send="/usr/local/zabbix_agentd/bin/zabbix_sender -c /usr/local/etc/zabbix_agentd.conf"
zbx_server="xibbaz"

4. adding UserParameter into /usr/local/etc/zabbix_agentd.conf

UserParameter=apache[*],/home/zabbix/bin/zabbix_apache.sh $1

5. Import the template from XML file >>> create graph.
but it's still not working so far.

Adding my firewall configuration



Please help me!

[steveboyson]

Ok, that requires some more questions. Are these assumptions correct?

- xentos
is the machine which apache webserver you want to monitor and where the script is supposed to run
- xibbaz
is the zabbix server

Can you please check if all items from the imported template are correctly imported? Check under "Configuration - Hosts - xentos" and look for items named "apache["

Is the script hooked into xentos' zabbix_agentd.conf file via "UserParameter=apache_stat[*]=/home/zabbix/bin/zabbix_apache.sh $1 $2 $3" entry?

Can you telnet to xibbaz port 10051 from xentos?

What does the zabbix_server.log say? Any errors?

[sccuser]

Dear steveboyson,

Exactly!
xentos = is the machine where apache running
xibbaz = zabbix server

I'm posting again the screenshot of items.

Is the script hooked into xentos' zabbix_agentd.conf file via "UserParameter=apache_stat[*]=/home/zabbix/bin/zabbix_apache.sh $1 $2 $3" entry?

Yes it's already configured, but just $1 only, not including $2 $3 as the above.

Yes, it's possible to ping tcp 10051 from xentos to zibbax



The zabbix_server.log

2551:20140304:151520.705 cannot send list of active checks to [10.x.x.x]: host [xentos] not found
2572:20140304:151538.638 cannot process step "Log in" of web scenario "Zabbix frontend" on host "Zabbix server": error in step variables "{sid}=regex:name="sid" value="([0-9a-z]{16})"": cannot extract the value of "{sid}" from response
2551:20140304:151720.794 cannot send list of active checks to [10.x.x.x]: host [xentos] not found
2572:20140304:151838.795 cannot process step "Log in" of web scenario "Zabbix frontend" on host "Zabbix server": error in step variables "{sid}=regex:name="sid" value="([0-9a-z]{16})"": cannot extract the value of "{sid}" from response
2553:20140304:151920.871 cannot send list of active checks to [10.x.x.x]: host [xentos] not found

10.x.x.x is the IP addresss of xentos

[steveboyson]

I think I found the error.

The host's name in Zabbix has nothing to do with the real hostname of the system.

I see that in zabbix your host is defined as "xentos...org" while in the script it tries to send data to a host called "xentos".

change the line in the script which reads "hostname -s" to just "hostname".

The "-s" strips the domain part from the hostname (which is fine for our environment but not yours).

[sccuser]

Dear steveboyson,

It's still persistent







From your XML template, I can see the Interval time is not configured, any clues?

[steveboyson]

What does

/usr/local/zabbix_agentd/bin/zabbix_sender -c /usr/local/etc/zabbix_agentd.conf -z xibbaz -s xentos.YOURDOMAIN.org -k 'apache[Stats_source]' -o FETCH

show?

The interval is only set for the item "Apache/workerScript" as it is the executed script. It has a defined interval of 300 seconds in the template.
All the other items are trapper items, that means the values gathered by the script get sent (pushed) to the zabbix server. They have no interval.

[sccuser]

Dear steveboyson,

Thanks for the explanation, please check the output

[steveboyson]

That looks good, it accepted the value. You should see it under "Monitoring - Latest data".

Now the script should be running ok, I think.

[sccuser]

Dear steveboyson,

It seems to be working just a half, at the time I run FETCH command.
The graph still not yet be displayed at all.






Should I change the "localhost" to xentos's IP address?

[steveboyson]

No, localhost is usually fine. Please check the permissions of the file /tmp/zbx_apache_localhost.txt (or the like). If it belongs to root, it wiull not work.

When you start the script as root, this file gets created. When zabbix wants to execute the script, it also tries to create that file but cannot due to the file already existing. Either remove it or give it "chown zabbix /tmp/zbx_apache_localhost.txt" so that the user under which the zabbix-agentd is running is able to overwite it.

What are the other "zabbix_send" commands in the script's output? Don't they make their way into zabbix?