VRTScavf (CVM) 7.4.2.2201 agent enhanced on AIX to handle EMC SRDF VxVM vxdg ERROR V-5-1-19179 Disk group AIXSRDF: import failed: SCSI-3 PR operation failed failures

Article: 100054524
Last Published: 2023-03-15
Ratings: 2 0
Product(s): InfoScale & Storage Foundation

Problem


The following issue is specific to AIX environments only, when the "dmp_cache_open" tunable is enabled on AIX in relation to Veritas SCSI-3 disk based fencing.
 

# vxdmpadm gettune dmp_cache_open
            Tunable               Current Value  Default Value
------------------------------    -------------  -------------
dmp_cache_open                           on               on



EMC hardware replicated SRDF devices may fail to import the VxVM managed disk group following a SRDF role switch.

The issue does not impact other hardware replicated solution such as Hitachi TrueCopy and other array vendor solutions. The issue is specific soley to EMC SRDF devices only,

The issue impacts both EMC PowerPath and Veritas DMP.

 

Error Message

 

# /usr/sbin/vxdg -s -o groupreserve=VCS -o clearreserve -cC -t import AIXSRDF
VxVM vxdg ERROR V-5-1-19179 Disk group AIXSRDF: import failed:
SCSI-3 PR operation failed

Please refer to system log for details.

 

Cause


The issue normally occurs following a system restart of the servers hosting the EMC srdf-r2 devices.

If a SRDF role switch is performed to promote the existing srdf-r2 devices to srdf-r1 devices, the import will fail to import as the revised device state has not been updated correctly on AIX.


SRDF ROLE SWITCH

in this instance, the EMC device group is named "AIXSRDF":

# symrdf -g AIXSRDF failover -establish -nopr

An RDF 'Failover' operation execution is
in progress for device group 'AIXSRDF'. Please wait...

Write Disable device(s) on SA at source (R1)..............Done.
Suspend RDF link(s).......................................Done.
Swap RDF Personality......................................Started.
Swap RDF Personality......................................Done.
Suspend RDF link(s).......................................Done.
Resume RDF link(s)........................................Started.
Resume RDF link(s)........................................Done.
Read/Write Enable device(s) on SA at target (R2)..........Done.

The RDF 'Failover' operation successfully executed for
device group 'AIXSRDF'.

 

To overcome this AIX specific issue, the newly promoted EMC srdf-r2 devices must be closed and rescanned by VxVM.

 

# vxdisk rm <dmpnode-names>
# vxdisk scandisks



To address this via Veritas Cluster Server, Veritas has introduced a series of new VCS CVMVolDg related attributes (ScanDisks = 1 & DGOptions) and a new vxdg import option -o usereplicatedev=only in connection with Veritas DMP.

 

Sample /etc/VRTSvcs/conf/config/main.cf
 

group srdf_SG (
        SystemList = { aixserver1-v3 = 0, aixserver2-v4 = 1 }
        AutoFailOver = 0
        Parallel = 1
        ClusterList = { PR_clust = 0, DR_clust = 1 }
        AutoStartList = { aixserver1-v3, aixserver2-v4 }
        )


        CFSMount childdir_cfs (
                Critical = 0
                MountPoint = "/parentdir/childdir"
                BlockDevice = "/dev/vx/dsk/AIXSRDF/vol2"
                MountOpt = rw
                )

        CFSMount parentdir_cfs (
                Critical = 0
                MountPoint = "/parentdir"
                BlockDevice = "/dev/vx/dsk/AIXSRDF/vol1"
                MountOpt = rw
                )

        CVMVolDg srdf_dg (
                CVMDiskGroup = AIXSRDF
                CVMVolume = { vol1, vol2 }
                CVMActivation = sw
                CVMDeportOnOffline = 1
                ClearClone = 1
                DGOptions = { "-t -o usereplicatedev=only" }
                ScanDisks = 1

                )

        requires group cvm online local firm
        childdir_cfs requires parentdir_cfs
        childdir_cfs requires srdf_dg
        parentdir_cfs requires srdf_dg

 

Solution

 

Veritas DMP requires the following Private hot-fixes to be installed for both packages, VRTSvxvm and VRTScavf.
 

VRTSvxvm (Veritas Volume Manager) 7.4.2 Hot Fix 2601

VRTScavf hot-fix 7.4.2.2102


The new DMP vxdg option "-o usereplicatedev=only" is now required for Hardware Replication solutions like EMC SRDF, Hitachi TrueCopy etc.

This new option ensures only Hardware Replicated devices are selected during the LUN selection process for the import operation.

 

CVMTypes.cf updated following the VRTScavf hot-fix install

 

Sample: /etc/VRTSvcs/conf/config/CVMTypes.cf

 

type CVMVolDg (
        static keylist RegList = { CVMActivation, CVMVolume, NodeList }
        static int ActionTimeout = 300
        static keylist SupportedActions = { vxdctlenable }
        static int OnlineRetryLimit = 2
        static int OnlineTimeout = 400
        static int OnlineWaitLimit = 10
        static str ArgList[] = { CVMVolume, CVMActivation, CVMVolumeIoTest, CVMDGAction, CVMDeportOnOffline, State, CVMDeactivateOnOffline, ClearClone, NodeList, Sc
anDisks, DGOptions }

        str CVMDGAction
        keylist CVMVolume
        str CVMActivation
        keylist CVMVolumeIoTest
        int CVMDeportOnOffline
        int CVMDeactivateOnOffline
        int ClearClone
        temp int voldg_stat
        keylist DGOptions
        int ScanDisks

        keylist NodeList
)

 

Improved logging done by CVMVolDg, tells a story of events now and errors seen during the import process


# tail -36 /var/VRTSvcs/log/CVMVolDg_A.log

2022/11/18 10:09:27 VCS INFO V-16-10011-1091 CVMVolDg:srdf_dg:online:Importing the diskgroup - AIXSRDF
2022/11/18 10:09:27 VCS INFO V-16-20007-1113 CVMVolDg:srdf_dg:online:ScanDisks is set to [1], disks associated with shared dg AIXSRDF will be scanned before importing
2022/11/18 10:09:28 VCS INFO V-16-20007-1440 CVMVolDg:srdf_dg:online:Disk Access (DA) Names associated with shared dg AIXSRDF are [emc1_0862 emc1_0863 emc1_0864 emc1_0865 emc1_0866 emc1_0867 emc1_0868 ]
2022/11/18 10:09:28 VCS INFO V-16-20007-1114 CVMVolDg:srdf_dg:online:Selective device scanning enabled, scanning OS device paths [hdisk219,hdisk157,hdisk42,hdisk101,hdisk220,hdisk158,hdisk43,hdisk102,hdisk221,hdisk159,hdisk103,hdisk44,hdisk222,hdisk160,hdisk45,hdisk104,hdisk161,hdisk223,hdisk46,hdisk105,hdisk162,hdisk224,hdisk47,hdisk106,hdisk225,hdisk163,hdisk48,hdisk107] prior to importing shared AIXSRDF
2022/11/18 10:09:29 VCS INFO V-16-20007-1113 CVMVolDg:srdf_dg:online:ClearClone set for shared dg AIXSRDF to be imported with vxdg [ -c] options to clear udid_mismatch and clone flags
2022/11/18 10:09:29 VCS NOTICE V-16-20007-1113 CVMVolDg:srdf_dg:online:Special DGOptions: [-t -o usereplicatedev=only] set for shared DiskGroup [AIXSRDF]
2022/11/18 10:09:29 VCS WARNING V-16-20007-1013 CVMVolDg:srdf_dg:online:Diskgroups will be imported with reservations
2022/11/18 10:09:29 VCS INFO V-16-20007-1113 CVMVolDg:srdf_dg:online:Attempting shared import for DiskGroup [AIXSRDF] with vxdg [ -c -t -o usereplicatedev=only -o groupreserve="VCS" -s] options
2022/11/18 10:09:30 VCS WARNING V-16-20007-1001 CVMVolDg:srdf_dg:online:vxdg import failed for dg AIXSRDF - error - VxVM vxdg ERROR V-5-1-19179 Disk group AIXSRDF: import failed:
Disk in use by another cluster

Please refer to system log for details.
2022/11/18 10:09:30 VCS INFO V-16-20007-1001 CVMVolDg:srdf_dg:online:Trying again with options [ -c -t -o usereplicatedev=only -o groupreserve="VCS" -s -C] to clear hostid.
2022/11/18 10:09:32 VCS INFO V-16-20007-1001 CVMVolDg:srdf_dg:online:VxVM vxdg ERROR V-5-1-19179 Disk group AIXSRDF: import failed:
SCSI-3 PR operation failed

Please refer to system log for details.
2022/11/18 10:09:32 VCS INFO V-16-20007-1001 CVMVolDg:srdf_dg:online:Disk group AIXSRDF failed with: VxVM vxdg ERROR V-5-1-19179 Disk group AIXSRDF: import failed:
SCSI-3 PR operation failed

Please refer to system log for details.
2022/11/18 10:09:32 VCS INFO V-16-20007-1113 CVMVolDg:srdf_dg:online:ScanDisks is set to [1], running clearimport for disks associated with shared dg AIXSRDF prior to running vxdctl enable
2022/11/18 10:09:32 VCS WARNING V-16-20007-1111 CVMVolDg:srdf_dg:online:Running clearimport against disk(s): [emc1_0862 emc1_0863 emc1_0864 emc1_0865 emc1_0866 emc1_0867 emc1_0868 ]
2022/11/18 10:09:33 VCS INFO V-16-20007-1113 CVMVolDg:srdf_dg:online:ScanDisks is set to [1], removing disks associated with shared dg AIXSRDF prior to running vxdctl enable

2022/11/18 10:09:33 VCS WARNING V-16-20007-1111 CVMVolDg:srdf_dg:online:Removing disk(s): [emc1_0862 emc1_0863 emc1_0864 emc1_0865 emc1_0866 emc1_0867 emc1_0868 ]
2022/11/18 10:09:34 VCS WARNING V-16-20007-1002 CVMVolDg:srdf_dg:online:Shared Disk group AIXSRDF import using [ -c -t -o usereplicatedev=only -o groupreserve="VCS" -s -C] failed. Will try running vxdctl enable.
2022/11/18 10:09:35 VCS INFO V-16-20007-1113 CVMVolDg:srdf_dg:online:/usr/sbin/vxdctl enable performed
2022/11/18 10:09:35 VCS INFO V-16-20007-1113 CVMVolDg:srdf_dg:online:Attempting shared Disk group AIXSRDF import again following vxdctl enable using [ -c -t -o usereplicatedev=only -o groupreserve="VCS" -s -C -o noreonline].
2022/11/18 10:09:38 VCS NOTICE V-16-20007-1005 CVMVolDg:srdf_dg:online:Shared vxdg import (after vxdctl enable) succeeded for Disk Group AIXSRDF using [ -c -t -o usereplicatedev=only -o groupreserve="VCS" -s -C -o noreonline]
2022/11/18 10:09:38 VCS INFO V-16-20007-1009 CVMVolDg:srdf_dg:online:AIXSRDF      enabled,shared,cds   1662552090.2110.claix04-v1
2022/11/18 10:09:41 VCS INFO V-16-10011-1046 CVMVolDg:srdf_dg:online:resource srdf_dg is online

 

VCS will attempt to run "vxdisk rm" against the related DMPNODE names associated with the VxVM disk group being imported.
 

# vxdisk -eo alldgs list | grep AIXSRDF
emc1_0c8e    auto:cdsdisk   emc1_0862    AIXSRDF     online shared        hdisk56        srdf-r1
emc1_0c8f    auto:cdsdisk   emc1_0863    AIXSRDF     online shared        hdisk92        srdf-r1
emc1_0c90    auto:cdsdisk   emc1_0864    AIXSRDF     online shared        hdisk93        srdf-r1
emc1_0c91    auto:cdsdisk   emc1_0865    AIXSRDF     online shared        hdisk59        srdf-r1
emc1_0c92    auto:cdsdisk   emc1_0866    AIXSRDF     online shared        hdisk95        srdf-r1
emc1_0c93    auto:cdsdisk   emc1_0867    AIXSRDF     online shared        hdisk61        srdf-r1
emc1_0c94    auto:cdsdisk   emc1_0868    AIXSRDF     online shared        hdisk97        srdf-r1

 

Note: The Veritas private hot-fixes have only been tested with Veritas DMP. We do not have a supported solution for EMC PowerPath (PP) impacting EMC storage arrays with Veritas fencing on AIX only.

Linux and Solaris platforms are not impacted by the AIX specific issue with EMC storage. 

The new vxdg "-o usereplicatedev=only" option will only work with Veritas DMP as the multi-pathing driver, due to the extra value-add DMP extended attributes exposed.
 

Was this content helpful?