Have SuSE10 server running GroupWise 8 domain, post office and gwia. It is scheduled to be migrated to new server in near future and upgraded to 2012. Server has 1 processor and 3 GB of ram, sp3 of SuSE10 installed. GWPOA process constantly running high utilization on server, threads all look fine. What are some things I can check to try and find out what is causing the issue?
Go into the HTTP console for the POA ( or enable it if you didn't ) and you can look at the C/S threads - it will show active threads and how long they have been executing the last command and who. Sometimes a wonky client will spin the POA performing a lot of cursor create / destroy activity.
Its unusual to have high CPU load, as GW is terribly efficient and typically you will see a lot of wa% ( I/O wait ) as it waits for the disks to deliver data.
If you go into top press 1 ( gives you a break down by cores ) and H which shows threads. Once you press H you should see multiple entries in top for various processes, including the POA. So the critical thing to look for is this: is the high CPU load spread between a LOT of POA threads, or just one or two?
But basically you need to sniff around in the listings of worker threads and see if you can find one or a bunch that are spinning like mad.
Also how many clients do you have? How many in C/S mode? Any backup / archiving products that may be making this busy?
In article <email@example.com>, Jlewter wrote: > GWPOA process constantly running high utilization on server, threads all > look fine. What are some things I can check to try and find out what is > causing the issue?
In addition to Bob's suggestions, check the POA logs for errors. One source of busy are the POA trying to deal with problems that need some human intervention to properly fix. One classic one is the Expire/Reduce Messages job constantly running if it frees up enough to drop below the threshold that is then quickly passed again. So check space. Check the results of any regular GWChecks: - How long are they running? Has your system grown such that they are running most of the time? - Or has their timing been picked that puts them during the busiest times and the POA just idles when nobody is using it? - Are there errors that don't go away? I.e. The same error in every check? For this you need to look at the details of each individual error. http://www.konecnyad.ca/andyk/gwlogs.htm
In article <firstname.lastname@example.org>, Jlewter wrote: > Not sure what you mean by go into top press 1? > When you are in top, press the key "1" (the key to the left of the "2" key) and the 3rd line of top will split from being a composite of all CPUs to each CPU core as its own line. There are a whole pile of single character toggles for top that gets it to present different data, such as lowercase "c" to get the full command of the process, though that works best on a terminal bigger/wider than the default 25 lines of 80 characters. Some times things really are that dead simple and our complex minds revolt against it ;)
Sorry I totally misread the original post. I thought you meant within the webadmin console for the po. I was like where do you have options in there to hit 1 or so. Put my glasses on and now I see you said with top. My bad.