Autopath Rerouting to alternate path

[Hka_1]

Hi All,

We are receiving many number of AUTOPATH path switch events in syslog. and some times we are facing performance issue with server during path failover.Please find the below error and advice what need to be done.

Feb 11 09:34:57 vmunix: AUTOPATH : Path 0xbc310000 failed! Rerouting to alternate path
Feb 11 09:34:58 vmunix: AUTOPATH : Path 0xbc2f9500 failed! Rerouting to alternate path
Feb 11 09:34:58 vmunix: AUTOPATH : Path 0xbc31d200 failed! Rerouting to alternate path
Feb 11 09:34:58 vmunix: AUTOPATH : Path 0xbc350700 failed! Rerouting to alternate path
Feb 11 09:35:07 vmunix: AUTOPATH : Path 0xbc310000 recovered
Feb 11 09:35:07 vmunix: AUTOPATH : Path 0xbc2f9500 recovered
Feb 11 09:35:07 vmunix: AUTOPATH : Path 0xbc31d200 recovered
Feb 11 09:35:07 vmunix: AUTOPATH : Path 0xbc350700 recovered
Feb 12 09:31:15 vmunix: AUTOPATH : Path 0xbc351100 failed! Rerouting to alternate path
Feb 12 09:31:32 vmunix: AUTOPATH : Path 0xbc351100 recovered
Feb 12 17:22:36 vmunix: AUTOPATH : Path 0xbc2f9200 failed! Rerouting to alternate path
Feb 12 17:22:41 vmunix: AUTOPATH : Path 0xbc2f9200 recovered
Feb 12 17:24:26 vmunix: AUTOPATH : Path 0xbc2f9400 failed! Rerouting to alternate path
Feb 12 17:24:31 vmunix: AUTOPATH : Path 0xbc2f9400 recovered

[TTr]

You have momemntary disk timeouts. They last for a few seconds. There can be several causes of this.
1. Your disks are extremely busy and they time out. You need to check and adjust the PV timeout limit if needed. If the disks are busy you also need to add some i/o capacity (more disks and/or interfaces)
2. You have flaky hardware. Try to narrow down the problem. From the 0xbc?????? numbers you can find out the disk devices in /dev/(r)dsk. See if they happen to the same disk groups in the array (i/o contention), same fiber interface (fc cable or switch or server HBA issue).

Did the start of these correlate to any specific event? Did they start afetr adding more load to the server? Or did they suddenly start (hardware flaked out).

There is a possibility that these are caused by lack of patching. Check for autopath and other system patches.

[Hka_1]

there are about 200 LDEVs assigned to this host and if there is any issue with FC link the failover to next available path should happen for all LDEVs, but in this case that is not happening, only few luns are getting switched.

I checked Switch ports and they are perfect and not pushing any error logs.and HBAs are also not pushing any error.