How to troubleshoot why a DX UIM probe is consuming high CPU
Article ID: 124801
Updated On:
Products
DX Unified Infrastructure Management (Nimsoft / UIM)
CA Unified Infrastructure Management On-Premise (Nimsoft / UIM)
CA Unified Infrastructure Management SaaS (Nimsoft / UIM)
Issue/Introduction
- Customer guidance on identifying and troubleshooting UIM Probes that may be consuming high cpu.
Environment
- Any UIM release
Cause
Possible causes include:
===================
(1) Old Java/Probe release
(2) Hotfix available and needed
(3) Configuration issues, for example:
(a) Insufficient probe memory
(b) issues with existing configuration / setup
(b) issues with existing configuration / setup
(4) Additional probe-specific issues:
(a) logmon probe - use of REGEX
(b) nas probe - replication failures
(c) snmpcollector probe workload
(d) database monitoring probes like Oracle, sqlserver - checkpoint timeout errors
(b) nas probe - replication failures
(c) snmpcollector probe workload
(d) database monitoring probes like Oracle, sqlserver - checkpoint timeout errors
Resolution
1. Identify the probe responsible for high CPU (refer to the Additional Information section)
2. Older Java/Probe release:
- It could be possible a probe has been running for quite some time and over this period, the java release it was designed upon has become obsolete (or) is no longer the latest one (or) a newer release of probe is available
- Refer to the probe release notes to find if a newer release is available and compatible and if yes, upgrade the probe to the latest release.
3. Hot-fix available:
- It would a good idea to refer to the hot-fix index to find if there is a hotfix created for the probe, and if yes, check the probe prerequisites.
- Refer to the readme before upgrading the probe to the hotfix level. The link to the hot-fix index can be found Probes Hotfix Page
4. Configuration issue(s):
(a) Insufficient probe memory:
=======================
One of the possible causes could be insufficient probe memory, if there is enough free memory available, try increasing the probe's initial and maximum memory (found in the Probe configuration startup -> opt section)
(b) Issues with existing configuration / setup:
==================================
You could try to take a backup and delete the existing probe (please be aware there could be data loss, if unsure about the consequences, reach out to Support), remove the probe folder manually, deploy the same (or) newer release of probe if available (provided you have read the Probe release notes)
=======================
One of the possible causes could be insufficient probe memory, if there is enough free memory available, try increasing the probe's initial and maximum memory (found in the Probe configuration startup -> opt section)
(b) Issues with existing configuration / setup:
==================================
You could try to take a backup and delete the existing probe (please be aware there could be data loss, if unsure about the consequences, reach out to Support), remove the probe folder manually, deploy the same (or) newer release of probe if available (provided you have read the Probe release notes)
5. Additional Probe specific issue(s):
(a) logmon probe - use of regular expressions:
=====================================
This is specific to the logmon probe, using general syntax like * as match pattern would increase the CPU load, try using more specific expressions to meet your requirement. The below links could be very handy when using Regex expressions:
Regex Basics
Including / Excluding Expressions using Regex
(b) nas probe - replication failure:
==========================
This is specific to the nas probe, in a multiple nas environment with Replication enabled, Replication failures could cause the probe to consume high CPU, check the nas probes on the respective robots are activated, running error-free, on the same latest updated release
(c) snmpcollector probe:
===================
This is specific to the snmpcollector probe, check if any profile is not getting activated.
=====================================
This is specific to the logmon probe, using general syntax like * as match pattern would increase the CPU load, try using more specific expressions to meet your requirement. The below links could be very handy when using Regex expressions:
Regex Basics
Including / Excluding Expressions using Regex
(b) nas probe - replication failure:
==========================
This is specific to the nas probe, in a multiple nas environment with Replication enabled, Replication failures could cause the probe to consume high CPU, check the nas probes on the respective robots are activated, running error-free, on the same latest updated release
(c) snmpcollector probe:
===================
This is specific to the snmpcollector probe, check if any profile is not getting activated.
- Probably a lot of load on the end server, refer to the link below to identify if too many metrics are created:
- Check if the end device is certified, refer to the below link for the device support list:
- It would also be good to refer to the below link for snmpcollector probe known issues:
(d) database monitoring probes like Oracle, sqlserver - checkpoint timeout errors:
==============================================================
This is specific to database monitoring probes like Oracle, sqlserver probes, check if there are checkpoint timeout errors in the probe log files (check the log files with debug enabled, click here to get an understanding on how to enable debugging), click here for further understanding.
If the issue continues to exist and/or if the above-mentioned points are not applicable, please open a case with Broadcom Support.
Additional Information
Procedure to identify the probe responsible for high CPU:
================================================
- Note down the PID (stands for Process identifier) for the identified High CPU consuming process:
Windows environment:
===================
Launch Task Manager -> Under Processes tab -> sort by CPU (descending order) to identify the process -> enable PID column -> note down the PID
Linux/UNIX environment:
===================
Find out the PID for the process consuming high CPU using the utilities like top, ps -ef to identify the offender, you may need to work with the respective System Administrator to get the information.
===================
Launch Task Manager -> Under Processes tab -> sort by CPU (descending order) to identify the process -> enable PID column -> note down the PID
Linux/UNIX environment:
===================
Find out the PID for the process consuming high CPU using the utilities like top, ps -ef to identify the offender, you may need to work with the respective System Administrator to get the information.
- From the Infrastructure Manager (or) Admin console, map the PID identified from the above procedure to find out the responsible probe
Feedback
Yes
No
Powered by
