HPE Server monitoring using the SNMP template -

[tcweb]

Good day, I have been working with HPE to address the issue of why zabbix SNMP monitoring of HPE iLO fails after some time.
When it works, it works great. However, after about 12-24 hours, the HPE iLO stops responding to SNMP requests.

I opened a case with HPE, and submitted logs, wireshark captures, and examples of what we are doing...showing them that it was NOT related to zabbix, but that regular CLI SNMP queries were eventually failing as well.

I did finally get a response. I hope this helps some others who are seeing the same issue. Current iLO firmware version is 3.10. 3.11 is due out (maybe) in April. I presume that late Q2 or early Q3 we can expect 3.12 that has these iLO memory issues resolved.

=-=-=-=-=
The L3 Engineering review is completed, and the resolution was shared as below.

[L3 Problem Description]: SNMP v3 monitoring via Zabbix stops responding within a time frame of several hours to several days. The only temporary workaround is the reset the iLO, but eventually, the issue will return.

[L3 Solution Description]: The Issue was found to be related to a bug in memory management on the iLO, and a fix will be provided in the 3.12 ilo5 FW version.

=-=-=-=-=

In the mean time, I have switched to using the iLO by HTTP template, which uses the Redfish architecture. I don't believe this template collects as many metrics by default, but it's pretty good.

-Tom

[tcweb]

Update, supposedly the iLO SNMP memory leak has been fixed. I plan to test shortly.

Version: 3.13 (25 Apr 2025)

Enhancements

Support for SHA2 algorithm with public key authentication method of logging into HPE iLO SSH and is applicable to all security modes.

Fixes

Upgrade Requirement:
Recommended - HPE recommends users update to this version at their earliest convenience.

---

• Fixed an issue where an HPE Gen10 Server health LED blinks displayed different behavior between HPE iLO v3.01and HPE iLO v3.02 upon a disk failure.

• Fixed an issue where a Redfish event not sent properly when event subscriptions used mTLS.

• Fixed an issue where DHE-RSA ciphers are enabled with Disable Weak Ciphers and in High Security mode setting making HPE iLO vulnerable.

• Fixed an issue where HPE iLO 5 v3.10 returned snmpwalk timeout when monitored by Zabbix monitoring software.

• Fixed an issue where the get_email_details failed if the user domain name had a \ character.

• Fixed an issue where a certificate import failed on some servers with the SSL certificate could not be imported since the input certificate is not intended for this server error.

• Fixed an issue where HPE iLO6 System Information Storage tab displayed drives without any order.

• Fixed an issue with the drive discovery, where an incorrect reporting of a removal or inserted drive IML message was getting triggered.

• Fixed an issue with HPE Apollo server power supply redundancy state.

• Fixed an issue where m750 server blades go to a critical state and does not connect to HPE iLO.

[jeffm_uccu]

I hate to raise this from the dead, but has your iLO5 monitoring worked since that update by HPE?

We have a new to us Gen10 with an iLO5, and on the current 3.16 firmware, we are seeing the exact behavior you described.
Monitoring with Zabbix works great for 12-36 hours, then SNMP stops responding.
Resetting the iLO brings it right back for another 12-36 hours.

We have several iLO3s and iLO4s that all work great with the same template, so it seems like there is still some kind of memory issue on the iLO5 that causes the SNMP service to hang or something.