Sun System Handbook - ISO 3.4 June 2011 Internal/Partner Edition | |||
|
|
Solution Type Technical Instruction Sure Solution 1004951.1 : Sun[TM] Cluster 3.x: How to change SCSI JBOD disk with Solstice DiskSuite[TM] SDS // Solaris Volume Manager[TM] SVM
PreviouslyPublishedAs 206944 Description This technical instruction explains how to change a SCSI JBOD (Just a Bunch Of Disks) under Solstice DiskSuite[TM] SDS / Solaris Volume Manager[TM] SVM in Sun[TM] Cluster 3.X (D1000, D2, 3310 JBOD...) Steps to Follow You must use metadevices, DID instances and disk labels that are valid on your system. Do not simply cut and paste commands from this document! Think before you act.
In the examples the following configuration was used. # metastat -p -s nfsset nfsset/d10 -m nfsset/d11 metaset/d12 1 nfsset/d11 1 1 d49s0 nfsset/d12 1 1 d50s0 # scdidadm -l d50 50 node0:/dev/rdsk/c3t2d0 /dev/did/rdsk/d50 # metadb -s nfsset flags first blk block count a m luo r 16 8192 /dev/did/dsk/d49s7 a luo r 16 8192 /dev/did/dsk/d50s7 1.) Change the quorum device if it's on the broken disk. Check the status with # scstat -q Use scsetup to change the quorum device. From the main menu select 1. # scsetup 2.) Identify the broken metadevice and did_instance number. # metaset # metastat -s nfsset # metastat -p -s nfsset # scdidadm -l d50 3.) Preserve the disk label, if needed. When the disk is not readable # prtvtoc /dev/rdsk/c3t2d0s2 > /var/tmp/c3t2d0.vtoc 4.) Identify the failed disk drive's physical DID node0 and node 1. # scdidadm -o diskid -l d50 5.) Delete metareplicas on that disk # metadb -s nfsset -d /dev/did/dsk/d50s7 6.) If hot spare disk is "In Use" go to step 10. 7.) Detach the broken mirror. # metadetach -s nfsset -f d10 d12 8.) Preserve metadevice configuration # metastat -s nfsset -p > /var/tmp/meta.txt 9.) Delete metadevice # metaclear -s nfsset d12 10.) Remove from metaset # metaset -s nfsset -d /dev/did/dsk/d50 11.) Unconfigure disk node0 and node 1. # cfgadm -c unconfigure c3::dsk/c3t2d0 12.) Physically replace disk ! 13.) Configure disk node0 and node 1. # cfgadm -c configure c3::dsk/c3t2d0 14.) Update the DID pseudo device on all nodes that have access to the disk. # scdidadm -R d50 15.) Verify that the new diskid is in place on all nodes that have access to # scdidadm -o diskid -l d50 16.) Reconfigure the global devices on all nodes. # scgdevs 17.) Label the disk, if needed. # fmthard -s /var/tmp/c3t2d0.vtoc /dev/rdsk/c3t2d0s2 18.) Add disk back to metaset # metaset -s nfsset -a /dev/did/dsk/d50 19.) If hot spare disk is "In Use" go to 23. 20.) Check that the metadevice was deleted in step 9. And then re-create it. # metastat -s nfsset d12 # metainit -s nfsset d12 1 1 /dev/did/dsk/d50s0 21.) Re-attach the detached mirror. # metattach -s nfsset d10 d12 22.) Go to step 24. 23.) Re-attach the detached mirror. # metareplace -s nfsset -e d10 /dev/did/dsk/d50s0 24.) If you want you can move the quorum device back to the old disk location # scstat -q Use scsetup to change the quorum # scsetup 22.) Check that everything is ok # metastat -s nfsset # metadb -s nfsset Product Sun Cluster 3.1 Sun Cluster 3.0 Sun StorageTek D2 Array Sun StorageTek D1000 Array scsi, jbod, disk, replacement, suncluster, cluster, scdidadm Previously Published As 77375 Change History Date: 2008-01-08 User Name: 95826 Action: Update Canceled Comment: *** Restored Published Content *** canceling update to allow migration to IBIS Version: 0 Attachments This solution has no attachment |
||||||||||||
|