PerfStack does not populate metrics

Greetings All.

A few days ago our Perfstack stopped populating metrics.

If you wait 15+ minutes it will sometimes populate metrics.

I found the below thread for a similar problem. However, both of the links to suggested fix action return 404 error - file not found.

perfstack charts are extremely slow to load (if at all)

Other Solarwinds functions appear to work as expected. No obvious error messages when attempting to run PerfStack.

We have not yet opened a ticket with Solarwinds. Wanted to try the Thwack community before going to Support.

I welcome any insights from the community on possible cause / fix for this issue.

Thanks.

Find more posts tagged with

slowness

PerfStack

Accepted answers

ralexander

Good tip regarding the Config Wizard. It was going to be my next step whilst the system was quiet.

We have the main engine, plus three additional pollers. We let the pollers do all the work so there's no nodes assigned to the main engine.
Just ran the Config Wizard on the main engine - the PerfStack charts were still empty for the nodes on the pollers. I moved a node onto the main engine and the charts appeared. Moved the node back to the poller and charts stopped again.

I've ran the wizard on the pollers now and its fixed all the charting. I'm going to put it down to the Windows Updates causing a problem, and the Config Wizard being the fix.

All comments

robert777

Additional troubleshooting note: if I create a report to return metrics, like avg cpu, memory in use, response time, the report returns the expected metrics with no noticeable delay. It's just anything that uses the PerfStack functionality in the web console that 'spins forever' and does not display data.

ralexander

This started to happen for us over the past couple of days as well. All PerfStack-based charts are stuck loading like in the below image.
The only thing that tends to change in our environment are applied Windows Updates, which was my first thought. Running 2024.4 here. 2024.4.1 is available but we've not yet installed.

robert777

Thanks, I appreciate the reply.

I'm guessing the first proposed fix action from the help desk would be "run the configuration wizard". For me the answer is no. I am fairly new on a multi-person team, working through some licensing glitches with the transition from legacy to Solarwinds Observability Enterprise Scale licensing, a couple of CVEs* with patches to load, and an upgrade in the works.

Beginning to make the point with my team that running the configuration wizard should be done on a regular basis. Hopefully once we work through some of the other fires we can give more attention to the Perfstack issue.

Btw - if you haven't run the Config Wizard on your SW system lately I would recommend that as a part of regular maintenance. I've seen multiple Thwack posts that Config Wizard should be run after loading Windows server patches

Cheers.

* CVE - "Common Vulnerabilities and Exposures" - security stuff

ralexander

robert777

Cool!

Blazing335

Sounds like a solid troubleshooting process! It’s interesting how moving a node to the main engine made the charts appear—definitely a good clue that the pollers needed a refresh. Windows Updates causing weird issues wouldn’t be surprising either. Glad running the Config Wizard on the pollers fixed everything! Thanks for sharing your experience—this will definitely help others facing the same issue.