Unsolved
This post is more than 5 years old
3 Posts
0
1193
problems with multipath on the RHEL 5
hello everyone
I installed a new system in two red hat qlogic HBA 2342 and gave a lun from a CX4-960 and set the multipath, it worked fine without problems. that occurs when I remove one of the fibers from the host to test the failover takes 30 seconds to take the other path lun, someone has seen this kind of situation?
the multipath instalations occurs without problems.
thanks
gc_emc
45 Posts
0
April 22nd, 2011 06:00
Hi,
It might be worth reviewing the "EMC® Host Connectivity with QLogic Fibre Channel and iSCSI Host Bus Adapters
(HBAs) and Fibre Channel over Ethernet Converged Network Adapters (CNAs) for the Linux Environment "guide.
You can find this on powerlink.emc.com - just search for the guide title above.
Check the QLogic BIOS settings for Fibre Channel HBAs and ensure they are at the recommended settings.
Also refer to the "EMC Host Connectivity Guide for Linux" -if you are using MPIO - go to page 218 - MPIO configuration details for RHEL 5
Rdgs
Gearóid
SKT2
1.3K Posts
0
May 21st, 2011 14:00
Sounds like that is an expected behaviur. Not much people prefere the fail over to happen for a small time out. it gives the an option to stay on the current path and if the problem lasts more than 30 seconds the fail over will be triggered. # cat /proc/scsi/qla2xxx/2|grep down Link down Timeout = 030 Port down retry = 030 Even in Hp-UX with LVM there is a PV time out which is ideally set to 90 seconds for both Clar and SYM. In Linux LVM there is not such an LVM time out i had seen, but there is a device level time out in OS level which is default 30 seconds. # cat /sys/block/sdz/device/timeout 30 # cat /sys/block/sda/device/timeout 30