Cluster Volume Manager (CVM) join fails if vxdiskadm is used to prevent multipathing of all paths

book

Article ID: 100005399

calendar_today

Updated On:

Description

Error Message

Feb 21 13:19:10 sys2 daemon:notice Had[389322]: VCS ERROR V-16-1-1005 (sys2) CVMCluster:???:monitor:node
 - state: out of cluster reason: Cannot find disk on slave node: retry to add a node failed
 

vxconfigd logged the following error messages.

VxVM vxconfigd NOTICE V-5-1-7900 CVM_VOLD_CONFIG command received
VxVM vxconfigd NOTICE V-5-1-7899 CVM_VOLD_CHANGE command received
VxVM vxconfigd WARNING V-5-1-8222 slave: missing disk 1257674215.116.c7prd04
VxVM vxconfigd WARNING V-5-1-7830 cannot find disk 1257674215.116.c7prd04
VxVM vxconfigd ERROR V-5-1-11092 cleanup_client: (Cannot find disk on slave node) 222
VxVM vxconfigd ERROR V-5-1-11467 kernel_fail_join() :           Reconfiguration interrupted: Reason is retry to add a node failed (13, 0)
VxVM vxconfigd NOTICE V-5-1-7901 CVM_VOLD_STOP command received

Error code 222 is defined as:

#define VE_CLUSTER_DB           222     /* Cannot find disk on slave node */
 

Cause

Customer used vxdiskadm to prevent multipathing of all paths because customer preferred to use AIX MPIO instead of Veritas Dynamic Multipathing (VxDMP).

Please note the use of MPIO in Veritas Volume Manager (VxVM) environment is not recommended by Symantec.   Please refer to the Veritas Storage Foundation and High Availability Solutions Hardware TechnNote (SymWISE article TECH47728) for details of the supportability of AIX MPIO.

output of vxdmpadm listenclosure all
-------------------------------------

ENCLR_NAME        ENCLR_TYPE     ENCLR_SNO      STATUS       ARRAY_TYPE     LUN_COUNT
=====================================================================================
other_disks       OTHER_DISKS    OTHER_DISKS    CONNECTED    OTHER_DISKS           70
 

As all paths were prevented from multipathing through DMP, the udid will contain disk name, and it would cause problem on cvm join as the udid becomes different for same disk.

$ grep udid vxdisk_list_hdisk67
udid:      IBM%5F2107900%5FOTHER%5FDISKS%5Fc7prd04%5F%2Fdev%2Frhdisk67
                                                                ^^^^^^
 
$ grep udid vxdisk_list_hdisk70
udid:      IBM%5F2107900%5FOTHER%5FDISKS%5Fc7prd03%5F%2Fdev%2Frhdisk70
                                                                ^^^^^^

Resolution

If customer insists on using AIX MPIO (which is not recommended by Veritas), then DMP should at least be configured as JBOD, so udid will get disk serial number instead of disk name.

Steps to change to JBOD mode:

(1) exclude the Array Support Library (ASL) for the corresponding diskarray.   Please note that if the disks are under the control of AIX MPIO, the VxDMP ASL should not be used.
# vxddladm excludearray libname=
 
(2) add array as jbod
# vxddladm addjbod vid= pid=

(3) allow dmp
# vxdiskadm
17     Allow multipathing/Unsuppress devices from VxVM's view
6      Allow multipathing of a disk by VxVM
Enter a pathname or pattern : [,all,list,list-exclude,q,?] all
 

output of vxdmpadm listenclosure all
---------------------------------------------------

ENCLR_NAME        ENCLR_TYPE     ENCLR_SNO      STATUS       ARRAY_TYPE     LUN_COUNT
=====================================================================================
disk              Disk           DISKS          CONNECTED          Disk            70

 

 

Applies To

AIX 5.3 / SFCFS5.0MP3 / MPIO / DS8K

Issue/Introduction

CVM join failed with below errors, but all shared disks are presented on both nodes.