vmware running platform v.3 cpu 100%

[Ray Mullen]

Hi we have just upgraded from platform 5.2.1 to 5.3 rev 3 and our ie crashes every night due to cpu usage running at 100%.

The engine runs fine all day until 8.00 pm every night when a ftp job kicks off and picks up a file and sends to 2 receiving threads. after about half way through the cpu goes up to 100% and box becomes totally unresponsive.

I’m not too familiar with vmware but the ops division have given me the specification and the box apparantly has 3.00 gig processor and 2 gig of ram and is running on windows 2003 service pack 1.

This tread ran fine on standalone box but it was platform 5.2 and we upgraded both software and hardware at same time.

Any help on this matter would be greatly appreciated. I have even moved the thread to a new process and still to no avail as the CPU is causing the box to be unresponsive when this ftp thread runs.

Topic

[Michael Hertel]

Just to possibly comfort you,

I’ve seen this behavior in the past with ftp threads taking over the process and not letting go. SNA threads used to do this too.

We’ve isolated them in their own process like you have.

We are on AIX so I assume the processes are playing nice in the sandbox.

I don’t beleive windows has a supervisor to keep errant processes in line but I could be wrong.

I don’t know if this has been fixed in other releases.

Stupid suggestion: Go back to 5.2.1 or change to Unix. 😯

I know this post was not of much help.

[Ray Mullen]

Thanks Michael, unfortunately when we went from 5.2.1 to 5.3 we also went to a vmware environment , so we dont know if its 5.3 or the vmware.  hopefully we will get it sorted soon. Wouldnt like to go back to 5.2.1 as it would be a step backwards, its hard enought to get the time to get an upgrade done.

thanks again

Ray

[Alice Kazin]

Did you ever find out if your issue was related to VM ware or your Cloverleaf client?  Do you still use VMware?

[Russ Ross]

I don’t know if your vmware environment is the same as the one I saw a demo on a while back but what you are describing reminds me of a configuration that was part of the demo I saw.

The virtual machine software I saw was dynamic and could be configured to allow a server like an FTP server very granular CPU usage in increments of one-tenth of a CPU up to everything available in the pool.

If your vmware has this type of configurablity, check that all the VM servers are restircted to prevent one of them from consuming the entire CPU pool.

It could be another server not obviously apparent in the VM pool that is consuming everything.

My gut feeling would make mean lean towards investigating the vmware configuration as how to solve your challenge.

By the way I was also told in our demo that not everything is capable of being dynamicaly allocated in a VM environment and SNA was one of those things that caught my attention as requiring a static allocation.

I’m not saying this is true but it is what I recall hearing.

All in all I sense that virtual machines are the wave of the futre.

I currently use statically allocated LPARs on AIX servers as my virtual machine solution and have found many benifits I never had previously, especially with upgrades.

Russ Ross
RussRoss318@gmail.com

[Ray Mullen]

Hi Alice and Russ,

Sorry about the late reply, To be honest I dont know much about vmware, we have an ops divison that sets up our servers e.t.c

and they would deal with all the vmware stuff.

What the guy’s did was change the configuration to a different version of vmware for the time being this apparantly dosent share the cpu resources and this solved the problem for a while. but on the odd occasion the ftp would cause it to go again.

Three weeks ago the guys gave the ie another cpu and the engine has been running like a dream ever since. They plan to move it back to the esx vmware environment pretty soon and hopefully we will be ok from now on.

[Vincent Tran]

I’m not sure if this is related…

Last month I noticed our Cloverleaf vmware instance shut down sometimes during the night. The image was stored on a Raid 5 disk configuration.

Due to some hard drive errors, I actually had to reformat/reinstall everything onto the machine. This time I used logical volume mounts instead of Raid and the Cloverleaf vmware instance runs now without shutting down every night.

– Vincent

[Ray Mullen]

thanks vincent, i’ll mention to our support guys

ray

[Author]

Replies