Currently working on issue with HPUX 11.31. I have a VMAX with device B60 presented to same host (HPUX 11.31) in two different paths. The devices are seen as c1t1d3 and c2t1d3 in host side. When I have created a volume (HP volume manager), I can add c1t1d3 as primary and c2t1d3 as alternate link (using pvlinks)
Whenever I try to vgreduce to remove the primary link and reduce the path 1 instead of 2, the vgreduce command works fin but "vgcfgbackup" command hangs and times out with "IO Error on Physical volume device /dev/dsk/rdsk/c2t1d3"
Does anyone experience this issue?
I have upgraded to new drivers for HBA based on HP call. But it did not resolve the issue. Our EMC resident engineer, filed a call with EMC. I am following up with HP. We had checked and rechecked the SAN and port settings. we do not see any issues. Any type of access to alternate path c2t1d3 takes too much time and timesout.
The underlying problem is the behaviour of scsimgr with "agile devices". to make use of legacy device paths, the leg_mpath_enable flag need to be set. (HP made it really convoulated in 11.31). When I set this flag properly, I could get pvlinks to work properly.
dynamox
11 Legend
•
20419 Posts
•
87439 Points
734
0
Posted June 23rd, 2011 14:00
i remember back in the days we used to set pvtimeout value to 180, worth a shot ?
pvchange -t 180 /dev/dsk/c2t1d1