vxconfigd daemon hangs when Veritas InfoScale Storage or Veritas InfoScale Enterprise is run on AIX7.2SP1 or any earlier version (3901325)
When Veritas InfoScale Storage or Veritas InfoScale Enterprise is run on AIX7.2SP1 (or any earlier version), a unique product hang issue occurs because of the
vxconfigd process getting hung. The vxdisk command, vxdg command, and vxprint command do not respond. In a clustered environment, if this issue hits on any one of the nodes, the commands may get hung on entire cluster nodes as well. The vxconfigd process that call IBM ODM-related interface functions hangs with the following stack trace
(0)> f 2658
pvthread+0A6200 STACK:
Use current context [F000000030019600] of cpu 4
[0065D0B0]kwpar_getmrc+000050 (F1000A0150520420 [??])
[0013EB30]sigprocmask+0004D0 (??, ??, ??)
[00003938]syscall+000230 ()
[D030635C]system+00019C (??)
[D06A6968]clr_disk_odm_owner+0000E8 (??, ??, ??)
[100E3978]devintf_clr_disk_odm_owner+000118 (00280018)
[100E3AE4]devintf_unstamp_odm+0000C4 (30337DD4)
[100CC904]req_disk_op+000264 (30246040, 30189328)
[10048DA8]request_loop+0011E8 ()
[10002BE0]main+001620 (00000002, 2FF22C50)
[10000208]__start+000068 ()
[kdb_read_mem] no real storage @ FFFFFFFFFFF9240
(0)>
This occurs because of an issue with the IBM AIX's Object Data Manager (ODM).
Workaround : IBM has fixed this issue by APAR IV80412. All the InfoScale produce nodes are required to have this fix, either by installing IBM's APAR IV80412, or by upgrading the system to AIX7.2 SP2, which already has this APAR fix included in it.