-1

The company I work for uses a virtual server for computers in the office and workshop to connect to and do work on. At one time, there are approximately 25 terminals connected to this server.

Last week, we had a brownout that lasted for about 30 seconds; after the brownout, we noticed the server was running slow. After further inspection, we found that instances of Chrome for 2-3 users at a time were each hogging up as much as 30% of the entire server's resources. This kind of usage happens even when they have <5 tabs open on common sites.

To try to alleviate the problem, I have disabled hardware acceleration and no longer allow Chrome to run apps in the background after it is closed for everyone. According to scans, no malware seems to be present either. However, the problem still persists.

What could be causing this to happen?

For reference, we have a dual-socket Xeon E5-2620 v3 @ 2.4 GHz, each with 14 virtual processors.

  • Does this occur for all or only some users? What OS is the server running? – I say Reinstate Monica Aug 11 '17 at 17:29
  • We're running Windows Server 2012 R2, and it happens for all users. – Michael Lilley Aug 11 '17 at 17:31
  • analyze the CPU usage with WPR/WPA similar to [this topic, but here expand the Chrome.exe](https://superuser.com/a/1164299/174557) instead of System, which has the high **Weight %** in the graph/Tbable. – magicandre1981 Aug 12 '17 at 07:39
  • I missed a step, you need to add the [chrome symbol server](https://www.chromium.org/developers/how-tos/debugging-on-windows) to [WPA symbol settings](https://docs.microsoft.com/de-de/windows-hardware/test/wpt/loading-symbols): – magicandre1981 Aug 12 '17 at 15:55

1 Answers1

0

The problem seems to have fixed itself, we're back at normal usage. We will look more into it and see if we can track down the source (and I will edit my answer if we find it), but for now I am closing this question.

Thank you to everyone who provided suggestions.