Error Message

vxconfigd dumps core. Restart of vxconfigd results in SEGV core.  The core can occur with execution of:

vxdctl enable

vxdisk scandisks

Stack traces of the core appear like:

core 'core' of 7535:    /sbin/vxconfigd -k -x syslog
-----------------  lwp# 1 / thread# 1  --------------------
001afd0c ddl_change_pwwn_of_path_instr (5974d8, 609100, 0, 97a9a5ce, 1d801a0, 5974d8) + 118
001927f0 ddl_migration_devlist_found (597568, 4000, fffffffe, fffffffd, 0, 34bc50) + 810
00191c14 ddl_check_migration_of_devices (5ed6b0, 5e8f88, ffbfe4b4, 596118, 47c00, 0) + b8
0019179c ddl_reconfigure_all (47c00, 36f280, 34bc50, 3939fc, 0, 395f98) + 390
0016e050 ddl_find_devices_in_system (393a10, 0, 0, 0, 36f284, 34bc50) + 43c
0010b69c find_devices_in_system (2, ffbff268, 0, 0, 0, 36f000) + 38
0010ab0c mode_set (2, ffbff4b4, 11, 0, 0, 0) + 1a8
00100364 setup_mode (2, 2ecc00, 0, 36d800, a39, 2ecc00) + 24
001002b4 startup  (36d800, 3b2c00, 36d800, 2ecc00, 36f000, 1d6d) + 2d4
000ffaac main     (36d800, 3cf400, 38d400, 2ec000, ffbff814, 0) + fb8
00041cf0 _start   (0, 0, 0, 0, 0, 0) + 108

-----------

 core file = /tmp/core -- program ``/sbin/vxconfigd'' on platform SUNW,Sun-Fire-
880
SIGSEGV: Segmentation Fault
ddl_migration_devlist_found+0x3a4(8f0c70, 3fe9bc, 3e95e4, 0, 8f1138, 0)
ddl_check_migration_of_devices+0xb4(8f0c70, 6fd2b8, ffbfe904, b09f70, 1, 1)
ddl_reconfigure_all+0x2f4(4424e0, 40ebe0, 40ebdc, 40db84, 3fe9bc, 42cbd0)
ddl_find_devices_in_system+0x448(0, 40c4b8, 0, 3e95e4, 0, 0)
find_devices_in_system+0x60(3, ffbff1d0, 0, 0, 0, 0)
mode_set+0x2d8(3, ffbff978, 0, 0, 0, 0)
setup_mode+0x18(3, 39c328, 4426a4, 0, 626c6500, 626c6500)
startup+0x338(0, ffbffb84, 39bc1c, 3e95e4, 48edc7, 5)
main+0x1a18(4, ffbffb84, ffbffb98, 3f0400, 0, 0)
_start+0xb8(0, 0, 0, 0, 0, 0)

-------------- 

 vxconfigd dumped core in DDL (device discovery layer) code upon detecting multiple failovermodes for the same DMP meta-node.

After increasing the debug level:

#vxconfigd -k -x 9 -x syslog

or

#vxdctl debug 9

 the vxconfigd.log (syslog for the architecture) will show similar messages

09/17 16:28:14:  VxVM vxconfigd DEBUG V-5-1-14475 Enclosure is CK200071000000:52:DGC:CLR-A/PF:EMC_CLARiiON:0
09/17 16:28:14:  VxVM vxconfigd DEBUG V-5-1-14475 Enclosure is CK200071000000:52:DGC:ALUA:EMC_CLARiiON:0

The same enclosure is being seen as both ALUA (in this case mode 4 for Clariion) and A/PF (also in this case mode 1 for Clariion).
Note: the Enclosure # is identical.

This can occur if the array had existing LUNS configured as AP/F.  Subsequently, some ALUA lun paths were added to the configuration.

Another possibility is a known (Primus solution emc246406) issue with EMC naviagent with Clariion :

"Failover Mode setting on CX4 Series array is unexpectedly reset from one value to another"

 This discrepancies on the failover mode of the array setting can be observed in ASLDebug output: (ATYPE)

scsi inquiry                           : /dev/rdsk/c4t5006016139A03A83d2s2
Vendor Identification                  : DGC    
Product Identification                 : RAID 5         
Revision                               : 0326
Serial Number                          : F200008990CL
libvxCLARiiON.so:claim_device()        : CLAIMED
ATYPE                                  : ALUA  -----------<<< Note Array Type
scsi inquiry                           : /dev/rdsk/c5t5006016B39A03A83d2s2
Vendor Identification                  : DGC    
Product Identification                 : RAID 5         
Revision                               : 0326
Serial Number                          : F200008990CL
LibvxCLARiiON.so:claim_device()        : CLAIMED
ATYPE                                  : CLR-A/PF -----------<<< Note Array Type
ARRAY_VOLUME_ID                        : 2034
 

Veritas DMP does not support devices in the same enclosure that are configured in different array type modes, i.e., one device is configured as ALUA and another one is configured as Active/Passive (A/P). Change all LUN paths to either A-P/F or ALUA and reboot the server.  If the LUN configuration issue is corrected but the system is not booted, vxconfigd would still dump core. So, a reboot is a must.

We have documented this limitation in the Veritas DMP 6.0.1 release notes (https://docs.infoscale.com/):

SOFTWARE LIMITATIONS

Veritas is working on enhancing the product in future releases (SF 6.1 or later release) so that vxconfigd does not crash in situations where LUNs from the same array enclosure have mixed mode settings.
 

Applies To

Active / Passive storage capable of mixed path failover modes.

persistent vxconfigd core after failover mode change in one or more paths to storage.

ETrack: 2255182