[B]Monitoring hp servers with Zabbix[/B]

[sga]

Good afternoon, all right?

I'm trying to do the monitoring of two HP servers, but I'm not getting any way. The servers of my client it does not have HP iLO and all I found in my research said that the templates using this HP iLO to reading the data. Someone monitors HP servers without using HP iLO or type tools? Can you help me in this matter?

Thank you and I'm sorry for bad English.

[Linwood]

There are a number of options.

If they run windows or linux or an OS that supports SNMP, you can monitor some aspects of them via SNMP.

If they run Windows (maybe linux not sure) you can install the HP Insight agent (you do not need the servers) and set them up to respond to SNMP, and those provide a LOT of visible information as well, more frankly than I think worth looking at. There are a few SNMP items for "overall" status you can poll to get indication of a problem, then look in the server (by human eye not zabbix) to find out what went wrong.

If they are running a specific application, that application might have ways to monitor it (e.g. database servers, web servers, etc.).

I guess you may need to narrow it down a bit what it is you want to monitor.

[sga]

There are a number of options.

If they run windows or linux or an OS that supports SNMP, you can monitor some aspects of them via SNMP.

If they run Windows (maybe linux not sure) you can install the HP Insight agent (you do not need the servers) and set them up to respond to SNMP, and those provide a LOT of visible information as well, more frankly than I think worth looking at. There are a few SNMP items for "overall" status you can poll to get indication of a problem, then look in the server (by human eye not zabbix) to find out what went wrong.

If they are running a specific application, that application might have ways to monitor it (e.g. database servers, web servers, etc.).

I guess you may need to narrow it down a bit what it is you want to monitor.

Good morning Linwood, how are you? I'm actually wanting to monitor their hardware, I forgot to specify the question forgive me. You monitors some there? Can you help me with this?

[Linwood]

I don't know how much good what I do will help, but you are welcome to it.

The first attachment is a simple template that has three items that are checked. Those are three OID's that are ancient, and I have not checked in some years to see if others are available, but collectively at least one of these tends to alert if anything at all goes wrong in the hardware monitoring of the HP server. To use it the Agent from HP must be installed, though it does not need to be monitored by a Insight server. It does need to be set up so it works with SNMP (It's been 3 years since I did an actual install so I will not try to remember how, but I do remember it was quite simple from the HP kits). On windows, never tried it on another OS.

This does not monitor the hardware, it monitors some rollup items that monitor hardware. The agents do offer SNMP OID's for all the individual hardware items I think, but frankly I never cared. If something went wrong, I wanted a tech to take a look, not evaluate from an email. So all they would do is hit the management page URL for the agents on the server and it shows what is wrong in detail.

And the alternative is to build a template with a huge number of LLD's for all the various pieces the agents monitor.

The fourth item is a trivial shell script that checks to see if the agents are even installed properly so they can respond. So I know which are the physical servers, attach this template directly (not let it be discovered), and this will trigger an error if the agents are not responding, the other three items will trigger if the agents respond with any anomaly. It's specific to an integer return, it is not general purpose, but could be made so easily.

#!/usr/bin/env bash
#
# Call: $1 = node address
#       #2 = community
#       #3 = OID
#
# Return: 0 = found
#         1 = not found, timeout or other error
#
x=$(/usr/bin/snmpwalk -v2c -c $2 $1 $3 2> /dev/null )
if [[ $? != 0 ]]
then
    echo 1
    exit
fi
if [[ $x == *"INTEGER"* ]]
then
   echo 0
else
   echo 1
fi

I haven't done a lot of proliant exploration in the last couple years, so if you find this does not work, or you improve it, let me know.

[sga]

I don't know how much good what I do will help, but you are welcome to it.

The first attachment is a simple template that has three items that are checked. Those are three OID's that are ancient, and I have not checked in some years to see if others are available, but collectively at least one of these tends to alert if anything at all goes wrong in the hardware monitoring of the HP server. To use it the Agent from HP must be installed, though it does not need to be monitored by a Insight server. It does need to be set up so it works with SNMP (It's been 3 years since I did an actual install so I will not try to remember how, but I do remember it was quite simple from the HP kits). On windows, never tried it on another OS.

This does not monitor the hardware, it monitors some rollup items that monitor hardware. The agents do offer SNMP OID's for all the individual hardware items I think, but frankly I never cared. If something went wrong, I wanted a tech to take a look, not evaluate from an email. So all they would do is hit the management page URL for the agents on the server and it shows what is wrong in detail.

And the alternative is to build a template with a huge number of LLD's for all the various pieces the agents monitor.

The fourth item is a trivial shell script that checks to see if the agents are even installed properly so they can respond. So I know which are the physical servers, attach this template directly (not let it be discovered), and this will trigger an error if the agents are not responding, the other three items will trigger if the agents respond with any anomaly. It's specific to an integer return, it is not general purpose, but could be made so easily.

#!/usr/bin/env bash
#
# Call: $1 = node address
#       #2 = community
#       #3 = OID
#
# Return: 0 = found
#         1 = not found, timeout or other error
#
x=$(/usr/bin/snmpwalk -v2c -c $2 $1 $3 2> /dev/null )
if [[ $? != 0 ]]
then
    echo 1
    exit
fi
if [[ $x == *"INTEGER"* ]]
then
   echo 0
else
   echo 1
fi

I haven't done a lot of proliant exploration in the last couple years, so if you find this does not work, or you improve it, let me know.

Good afternoon all right
Then, the template does not matter, it gives error that depends on a trigger that is not found. Look mistake of import.

Import failed
Details

Created: Application "Proliant Agent Global Indicators" in "Template SNMP HP Proliant Agent".
Created: Item "Are proliant agents responding (1 = not responding)" in "Template SNMP HP Proliant Agent".
Created: Item "Proliant Array Health Indicator" in "Template SNMP HP Proliant Agent".
Created: Item "Proliant Overall Heath Indicator" in "Template SNMP HP Proliant Agent".
Created: Item "Proliant Health System Indicator" in "Template SNMP HP Proliant Agent".
Created: Trigger "Proliant Agents Not Responding" in "Template SNMP HP Proliant Agent".
Created: Trigger "Proliant Array Health Abnormal" in "Template SNMP HP Proliant Agent".
Created: Trigger "Overall Proliant Health Abnormal" in "Template SNMP HP Proliant Agent".
Created: Trigger "Proliant Health System Abnormal" in "Template SNMP HP Proliant Agent".
The trigger "Proliant Agents Not Responding" depends on the trigger "by Unavailable Ping Check" which does not exist.

Thanks again for helping me, and forgive the bad English.

[Linwood]

The trigger "Proliant Agents Not Responding" depends on the trigger "by Unavailable Ping Check" which does not exist.

Thanks again for helping me, and forgive the bad English.

Like a lot of template sharing, they often do not work in isolation. I had another that checks for pings and did not think to remove it. So you see how this stuff works, just edit (with any text editor) the XML file and find this:

            <dependencies>
                <dependency>
                    <name>Unavailable by Ping Check</name>
                    <expression>( {TRIGGER.VALUE}=0 and {Template Ping Checks:pingCheck[{HOST.CONN},{HOST.HOST}].last()}=0  ) or 
( {TRIGGER.VALUE}=1 and {Template Ping Checks:pingCheck[{HOST.CONN},{HOST.HOST}].min(10m)}=0  )</expression>
                </dependency>
            </dependencies>

Either remove this whole section, or change the trigger expression to one you have that means "server is down". The idea is not to give these errors if the server is down entirely.

[CloudKristen9]

Hii,

Follow below instructions for monitoring 2 Hp servers at a time, this may help you resolve problems

Edit the APT configuration file and add the following lines to the end of the file.

# vi /etc/apt/sources.list

deb http://ppa.launchpad.net/vvk/openipmi/ubuntu precise main
deb-src http://ppa.launchpad.net/vvk/openipmi/ubuntu precise main

Run the following commands to configure apt-get.

# apt-key adv --recv-keys --keyserver keyserver.ubuntu.com 5E3E08EF
# apt-get update


Note! These commands will address a bug in Zabbix IPMI monitoring. (libopenipmi)

# apt-get remove openipmi libopenipmi0
# apt-get clean
# apt-get install libsnmp-dev libpopt-dev libncurses5-dev libssl-dev
# apt-get install -t precise libopenipmi0
# apt-get install openipmi


Edit the zabbix_server.conf file and insert the line below.

# vi /usr/local/etc/zabbix_server.conf

StartIPMIPollers=5

Restart the zabbix server.

# /etc/init.d/zabbix-server restart


Run the following command to get a list of IPMI sensors available on your machine. (HP DL380)

# ipmitool -I lanplus -H 192.168.3.3 -U administrator -P kamisama123@ sensor