There is a server having an interesting problem (a few other had the problem too). We think that SAP takes almost all the paging, but we cannot say that 100%, because when this problem occurs, even a "ps -ef" won't run on the system because the command will hang!

During the problem (before a reboot, because a reboot fixes the problem), how can I do a dump, that the developers later can analyze?

So far I read that if a:

sysdumpstart -p

is executed the AIX will do a dump and reboot after it:

-p  Initiates a system dump and writes the results to the primary dump device.

Question: but is this enough? (the "sysdumpstart -p" command) Will it creates a dump that will store the SAP related infos too, to later debug?

12:root@SERVER:/root # sysdumpdev -l
primary              /dev/lg_dumplv
12:root@SERVER:/root # sysdumpdev -e
Estimated dump size in bytes: 4660710604
12:root@SERVER:/root # 

The lg_dumplv is 12288 MByte sized, so it should be enough.

After the reboot, I will find the dump files in the "/var/adm/ras/vmcore.x" ? Or there is an additional command to put the dump to the FS from the dump lv?

  • wouldn't the periodic nmon logging help you easily highlight the offending process ? – steve Dec 22 '16 at 21:46
sysdumpstart -p

it took about ~22 minutes to do this 4 GByte sized one. It automatically reboots after the dump! After reboot, save the dump from the dumpLV to a file.

smitty dump
Copy a system dump from a dump device to a file

trying to get a developer who can analyze the dump file :) opening software call.

How to force a system dump: https://www-01.ibm.com/support/docview.wss?uid=isg3T1019210

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.