System Administration
cancel
Showing results for 
Search instead for 
Did you mean: 

cfg_ksm_memreq failed message.

Christof Schoeman
Frequent Advisor

cfg_ksm_memreq failed message.

Hi

Yesterday, my customer added some EMC Clariion storage to his system and did a scsi scan, no problem.

Today, some more storage was added on another Clariion in the same fabric and he did a scsi scan again. The devices are now visible and accessible, but he periodically got the following message at the prompt:

"cfg_ksm_memreq failed"

The messages disappeared when the ssh session was closed, but he is still curious as to what caused them.

Any ideas?
16 REPLIES
Venkatesh BL
Honored Contributor

Re: cfg_ksm_memreq failed message.

What is the OS & patch version?

Was 'collect' running by any chance?

Can you post the syslogs?
Christof Schoeman
Frequent Advisor

Re: cfg_ksm_memreq failed message.

The OS version is 5.1B PK4 and yes, collect was running.

No messages in the logs around this time.
Venkatesh BL
Honored Contributor

Re: cfg_ksm_memreq failed message.

Are you able to see the devices thru "hwmgr -view devices" command?

Does the message pop up even if you stop 'collect'?
Christof Schoeman
Frequent Advisor

Re: cfg_ksm_memreq failed message.

Yes, all the devices are visible and accessible. The messages were only displayed on the one ssh session and disappeared when the ssh session was disconnected.

The customer is only curious as to what the messages actually meant. Is it something he needs to worry about?

Thanks for you help so far.
Venkatesh BL
Honored Contributor

Re: cfg_ksm_memreq failed message.

I suspect that the error was thrown out by 'collect'. It happens if collect is unable to get configuration info from the kernel. But, not sure about the connection with 'ssh'. Anyway, I don't think the customer need to worry about this.
Venkatesh BL
Honored Contributor

Re: cfg_ksm_memreq failed message.

btw, is this a cluster node?.

Also, did you check if you are able to playback the files generated by 'collect'? Especially, does the 'Disk Info' contain correct values?
Christof Schoeman
Frequent Advisor

Re: cfg_ksm_memreq failed message.

This is not a cluster node.

I'll have a look at the collect files, thanks.
Mulgund
Frequent Advisor

Re: cfg_ksm_memreq failed message.

It would be interesting to see for what reason
the memory allocation has failed.

Could be for not enough memory at disposal when scan scsi.

Could be a accumulated memory leak issue with hwmgr too ..

Need to check the relevant logs.

-Srinivas
Christof Schoeman
Frequent Advisor

Re: cfg_ksm_memreq failed message.

Mmmmm...

I'm now seeing this on another system too. Extra disks were added this morning. scsi scan picked them up fine.

I have two sessions open on the system, but only one of them is showing the message.

Collect files are also working fine.

Could not find anything out of the ordinary in the log files.

Anything in particular you would have looked at?
Mulgund
Frequent Advisor

Re: cfg_ksm_memreq failed message.

/usr/var/adm/syslog.dated/current/kern.log
Greg Yates
Valued Contributor

Re: cfg_ksm_memreq failed message.

Hi,

I did a quick look at this and it appears that there is a patch for this problem.

Please log a support call and we can see if you need this same fix.

Thanks.
Greg
Venkatesh BL
Honored Contributor

Re: cfg_ksm_memreq failed message.

I am wondering if 'collect' is throwing these errors. Can you stop collect and see if you still see these messages?
Greg Yates
Valued Contributor

Re: cfg_ksm_memreq failed message.

Yes, this is related to collect. There is a collect patch (I don't think it's on ITRC... yet) which corrects some other stuff (RAD/CPU fixes). That collect patch was given to another customer with this same problem and it solved this "cfg" problem as well.

Greg
Alet Barnard
Occasional Advisor

Re: cfg_ksm_memreq failed message.

Greg, can you send us the text files for the CSP.
Will this fix be included in PK5 - which should be coming out soon (June-ish)...

Regards
Alet
Greg Yates
Valued Contributor

Re: cfg_ksm_memreq failed message.

Alet,

The CSP I mentioned was to solve a CPU/RAD problem in collect

"...where the RADs are not serially numbered"

but it /apparently/ (correct me if I'm wrong, blv) solved this problem as well.

As for this fix being in the next aggregate patch kit, it doesn't look like it is.

Greg
Venkatesh BL
Honored Contributor

Re: cfg_ksm_memreq failed message.

Yes, the fix for CPU/RAD problems fixed this one too. Those fixes _will_ be available in PK5.