Originally posted by: SystemAdmin

I, i have the following config on a AIX 5.3 box:
2 FC adapters connect to a disk array throug a FC switch. I have MPIO configured in round_robin mode and each FC adapter "sees" 2 paths to a lun from the disk array.
I've been doing some tests and whenever i pull a FC cable on the disk array side it takes about 60 seconds for the MPIO driver to put that specific path offline and on that time i'm not getting any I/O done. During that 60 seconds interval my applications freeze.
Question: Is there any way to change this behaviour? Like changing the time that the driver takes to see that path is really offline...

Thanks for your help.

Originally posted by: SystemAdmin

Hi

We saw a similar issue with HDS storage and were advised by IBM to enable the Fast Fail option on the fiber cards, issue a "lsattr -El fscsi0" and look for the "fc_err_recov" line, if it says "delayed fail" it is still at the default.

To change it issue a "chdev -l fscsiXX -a fc_err_recov=fast_fail" for all fiber cards and reboot. For more info, search the IBM site for "fast fail" or I can send you more detailed info.

Cheers...Steve

Originally posted by: SystemAdmin

Hi Steve,

i've done the change you mentioned and the failover must faster. Another thing: I was able to change the parameters on the fly without having to reboot the box.
Thanks a lot for your help.

Originally posted by: SystemAdmin

Glad to hear it helped, the doc source for this parameter is http://publib.boulder.ibm.com/infocenter/pseries/v5r3/index.jsp and search for "fast_fail" you could also look at dynamic tracking if you have fibre ports being moved around.

Cheers...Steve

Originally posted by: sungkenkok

Hi! I'm ken! could you sent more detail of what you mentioned in mail to me! THKs! my email: kenkok@bj1860.net!
ASAP!