[Ace] Fwd: [OPS-PR] UPDATED: CC console opsl05 extremely sluggish

[Anthony Cuffe]

Not sure if you guys get these since he did an ops-pr.  Can someone investigate?
________________________________
From: moser at jlab.org <moser at jlab.org>
Sent: Saturday, June 18, 2022 10:40 AM
To: Anthony Cuffe <cuffe at jlab.org>
Subject: [OPS-PR] UPDATED: CC console opsl05 extremely sluggish

CC console opsl05 extremely sluggish<https://logbooks.jlab.org/entry/4004261>

  *   Edit<https://logbooks.jlab.org/entry/4004261/edit?destination=email/send>
  *   Delete<https://logbooks.jlab.org/entry/4004261/delete?destination=email/send>

Lognumber 4004261<https://logbooks.jlab.org/entry/4004261>. Submitted by moser<https://logbooks.jlab.org/user/moser> on Sat, 06/18/2022 - 10:14<https://logbooks.jlab.org/entries?start_date=1655558084&end_date=1655565284&book=ELOG>.

Last updated on Sat, 06/18/2022 - 11:54

PR (OPS)
Update this Record...<https://logbooks.jlab.org/pr/edit/1077293>
Needs Attention:
YES
Affected System:
CEBAF > Info Systems > Systems Administration
Responsible Group:
SysAdmin
There are 4 comments...
Problem Report Update<https://logbooks.jlab.org/comment/38466#comment-38466>
new

by moser<https://logbooks.jlab.org/user/moser> on Sat, 06/18/2022 - 14:36

w/minimal control screens up, trying to complete a do-log. two Firefox windows. I'm having to wait minutes for tabs to switch. When closing the BoxPS summary EDM screen, took ~30s to fully close. Wanted to check Ion Chamber calibration settings using iCalibrate, took over a minute to load in.

Problem Report Update<https://logbooks.jlab.org/comment/38460#comment-38460>
new

by moser<https://logbooks.jlab.org/user/moser> on Sat, 06/18/2022 - 11:45

Could it be CED?
<br>
I ***FINALLY*** got the couple tabs of Microsoft Outlook, Calendar, and Portal to close after several minutes, and the system was still extremely sluggish. Took ~10sec for an OTF quad screen to render. CED was still an open tab in Firefox, and I was logged into CED. Killing that tab took almost a minute.
<br>
Had another operator on a different console navigate to CED, and his system did not get bogged down and continued to work as expected. Same setup -- internal firefox, logged into CED...
<br>
Could it be something related to individual user accounts and CED? Like those with more permissions bog the database link down, slowing down their whole machine?
<br>
Before getting the CED tab finally closed, I was tempted to reboot the console AGAIN.

Problem Report Update<https://logbooks.jlab.org/comment/38459#comment-38459>
new

by moser<https://logbooks.jlab.org/user/moser> on Sat, 06/18/2022 - 11:22

The system is getting sluggish again. No adviewer or chrome-browse running. Firefox, internal mode, EDM, locks, etc. What is different now is I have Microsoft Office/Portal open in Firefox. Could this be the culprit?

htop after reboot...<https://logbooks.jlab.org/comment/38442#comment-38442>
new

by moser<https://logbooks.jlab.org/user/moser> on Sat, 06/18/2022 - 10:23

Systems responding much better now, but still curious as to why a single X-process is taking a whole core. I haven't brought up adviewer, have brought up Firefox, EDM, wish/tcl lock applications, jmenu etc.

[https://logbooks.jlab.org/files/styles/thumbnail/public/2022/06/4004261/Screenshot%20from%202022-06-18%2010-21-11.png]<https://logbooks.jlab.org/files/2022/06/4004261/Screenshot%20from%202022-06-18%2010-21-11.png>
Logbooks:       ELOG<https://logbooks.jlab.org/book/elog>
Entry Makers:   moser
References:     4004256 - The Wall Crashed in a New Way, Took >1 Reboot To Recover<https://logbooks.jlab.org/entry/4004256>

It was working OK when I first logged in, but currently as I type this into JELI I'm getting whole words ahead of the screen in terms of keyboard input and screen display.

I attended the RC meeting via Zoom using Firefox browser. Issues seem to begin around that time. Or maybe when I opened an adviewer window.

Figure 1: An htop sorted by CPU-%. I've only been logged in for ~2.5hrs, how are the adviewer applications open for 20+ hrs? I only opened a single adviewer instance -- what are all these other processes? One /usr/bin/X process is taking up a whole core? The screengrab to get Figure 1, btw, took ~5sec.

Submitting this log and rebooting. Linked log describes issues with the wall this morning. Perhaps they are related.

I'm beginning to think we need to examine our Control Room computing needs more closely, especially wrt to video-conferencing and accelerator viewer usage.

[https://logbooks.jlab.org/files/styles/plentybig/public/2022/06/4004261/22067125.1.png]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/ace/attachments/20220619/7b8cdd0c/attachment.html>