Veritas Cluster Server (VCS) DiskGroup Agent Solaris Sparc 7.3.1.104 includes new ScanDisks attributes to refresh VxVM and DMP attributes prior to importing disk group

Problem

Veritas Cluster Server (VCS) provides multiple replication agents to manage hardware replication functionality

• Prior to importing any hardware replicated or hardware clone device managed by Veritas Volume Manger it is vital the VxVM disk attributes are refreshed using vxdisk scandisks [device=<paths>]
  
• Some VCS replication agents (EMC SRDF, EMC MirrorView, Hitachi TrueCopy) already have additional intelligence built-in to perform selective device rescanning (vxdisk scandisks [device=<paths>]) following replication takeover
  

The VCS DiskGroup & CVMVolDg resources require “ClearClone = 1” to be defined for importing hardware replicated devices (vxdg -c import)
 

Figure 1 . Different Lunclasses

Error Message

Sample vxdg messages:

# vxdg -c import datadg
VxVM vxdg ERROR V-5-1-19179 Disk group datadg: import failed:
Device is a hardware mirror

The above error is normally seen with when replicated failover/takeover has NOT been done or the VxVM attributes were not been refreshed prior to importing the disk group

VxVM will display many different error messages depending on the condition encountered

• The “hardware mirror” error string is NOT always shown as part of the import error, even when the HARDWARE_MIRROR flag is set
  
  

Other vxdg error messages:

1.       Takeover NOT done: Disks in write-disabled state

# vxdg -c import datadg
VxVM vxdg ERROR V-5-1-19179 Disk group datadg: import failed:
No valid disk found containing disk group

Note: If a stale HOSTID has been detected, then use "vxdg -Cc" together.

# vxdg -c import datadg
VxVM vxdg ERROR V-5-1-19179 Disk group datadg: import failed:
Disk is in use by another host

Please refer to system log for details.

Solution

Veritas Cluster Server (VCS) handling of Hardware Clone & Replicated devices

DMP & VxVM Attributes

To increase the flexibility of managing future hardware technologies

The VCS DiskGroup agent will be taking over the responsibility of running the selective device scan (vxdisk scandisks device=<paths>) operations for replication agents which don’t have this capability.
 

VCSvcsag 7.3.1.104 for Solaris Sparc introduced NEW DiskGroup resource attribute called “ScanDisks”
 

VCS DiskGroup agent performing selective device scans
 

Figure 2 . Enhanced VCS design

Sample main.cf
 

DiskGroup tc_dg (
                      DiskGroup = datadg
                      ClearClone = 1
                      ScanDisks = 1
                      ForceImport = 0
                 )

Where “ScanDisks = 1” a selective device scan of all paths will be performed at a DiskGroup Level using the disk group name specified

To pro-actively assist customers in the field, for whom may not be aware of the new “ScanDisks” attribute upon its release.

The VCS DiskGroup agent will attempt to perform the selective device scan when the HARDWARE_MIRROR flag is detected by the vxdg import error message.

As the import error returned by the failed import may differ and does not always include the “hardware mirror” string, it is impossible for VCS to dynamically run the selective device scan in all cases.

Please contact Veritas Technical Support if you require the Solaris Sparc VRTSvcsag 7.3.1.104 Private Hot-fix
 

VRTSvcsag PATCH ID:7.3.1.104

 * INCIDENT NO:3985201               TRACKING ID:3986809

SYMPTOM:

In a hardware replicated environment, a disk group resource may fail to import when the HARDWARE_MIRROR flag is set

DESCRIPTION:

After the VCS hardware replication agent resource fails over control to the secondary site, the DiskGroup agent does not rescan all the required device paths in case of a multi-pathing configuration.

The vxdg import operation fails, as the hardware device characteristics for all the paths are not refreshed.

RESOLUTION:

This hotfix introduces of a new resource attribute for DiskGroup agent called ScanDisks. The ScanDisks attributes enables the user to perform a selective devices scan for all disk paths associated with a VxVM disk group. The VxVM and DMP disks attributes are refreshed before attempting to importing hardware clone or replicated devices. The default value of ScanDisks is 0, which indicates a selective device scan is not performed. Even when set 0, if the disk group fails with an error string containing HARDWARE MIRROR during the first disk group import attempt, the DiskGroup agent will then perform a selective device scan to increase of the chances of a successful import.

Sample resource configurations:

For Hardware Clone DiskGroups

DiskGroup tc_dg (
                  DiskGroup = datadg
                  DGOptions = "-o useclonedev=on -o updateid"
                  ScanDisks = 1
)

For Hardware Replicated DiskGroups
 

DiskGroup tc_dg (
                  DiskGroup = datadg
                  ClearClone = 1
                  DGOptions = "-o usereplicatedev=on"      (later changed =only)
                  ScanDisks = 1
)

Note: The VRTSvxvm patch level controls whether the import option is "-o usereplicatedev=on" or "-o usereplicatedev=only".
 

Sample VxVM & Hitachi TrueCopy (HTC) takeover commands:

1.] Display the VxVM extended attributes for Hardware Clone and Replicated type devices:

# vxdisk -px LIST_CLONE list

DEVICE               DG_NAME REPLICATED REPLICATED_TYPE HARDWARE_MIRROR LUN_SIZE  NR_DEVICE DEVICE_TYPE DDL_DEVICE_ATTR
hitachi_vspgx0_015d  datadg  yes        rep_sec         yes             10485760  -         -           hdprclm,fc,tc-svol
hitachi_vspgx0_015e  datadg  yes        rep_sec         yes             10485760  -         -           hdprclm,fc,tc-svol
hitachi_vspgx0_015f  datadg  yes        rep_sec         yes             10485760  -         -           hdprclm,fc,tc-svol
hitachi_vspgx0_0160  datadg  yes        rep_sec         yes             10485760  -         -           hdprclm,fc,tc-svol
hitachi_vspgx0_0161  datadg  yes        rep_sec         yes             10485760  -         -           hdprclm,fc,tc-svol

2.] Switch the HTC replication direction
 

# horctakeover -t 289  -g vg1 -I1
horctakeover : Swap-Takeover done.

3.] Refresh VxVM & DMP atrributes by running "vxdisk scandisks"
 

# vxdisk scandisks


4.] Review the refreshed VxVM extended attributes for Hardware Clone and Replicated type devices:
 

# vxdisk -px LIST_CLONE list

DEVICE               DG_NAME REPLICATED REPLICATED_TYPE HARDWARE_MIRROR LUN_SIZE  NR_DEVICE DEVICE_TYPE DDL_DEVICE_ATTR
hitachi_vspgx0_015d  datadg  yes        rep_pri         no              10485760  -         -           hdprclm,fc,tc-pvol
hitachi_vspgx0_015e  datadg  yes        rep_pri         no              10485760  -         -           hdprclm,fc,tc-pvol
hitachi_vspgx0_015f  datadg  yes        rep_pri         no              10485760  -         -           hdprclm,fc,tc-pvol
hitachi_vspgx0_0160  datadg  yes        rep_pri         no              10485760  -         -           hdprclm,fc,tc-pvol
hitachi_vspgx0_0161  datadg  yes        rep_pri         no              10485760  -         -           hdprclm,fc,tc-pvol
-           -

5.] Display the refreshed VxVM/DMP extended attributes:
 

# vxdisk -eo alldgs list
DEVICE       TYPE           DISK        GROUP        STATUS               OS_NATIVE_NAME   ATTR
hitachi_vspgx0_015d auto:cdsdisk   -            (datadg)    online thinrclm udid_mismatch c6t50060E80124FDD10d1s2              hdprclm fc tc-pvol
hitachi_vspgx0_015e auto:cdsdisk   -            (datadg)    online thinrclm udid_mismatch c6t50060E80124FDD10d2s2              hdprclm fc tc-pvol
hitachi_vspgx0_015f auto:cdsdisk   -            (datadg)    online thinrclm udid_mismatch c6t50060E80124FDD10d3s2             hdprclm fc tc-pvol
hitachi_vspgx0_0160 auto:cdsdisk   -            (datadg)    online thinrclm udid_mismatch c6t50060E80124FDD10d4s2              hdprclm fc tc-pvol
hitachi_vspgx0_0161 auto:cdsdisk   -            (datadg)    online thinrclm udid_mismatch c6t50060E80124FDD10d5s2              hdprclm fc tc-pvol
hitachi_vspgx0_0162 auto:none      -            -           online invalid thinrclm sdj              hdprclm fc