Disable write cache on SNO disks

Scope

SNO2U / 5U, SNO2.  Does not apply to SNOSC.

Started with GPFS 5.1.7.1

Problem

mmhealth node show shows failed state if drives has write cache enabled:

NATIVE_RAID FAILED 1 min. ago gnr_rg_failed(rg-sno-190153-A, rg-sno-190153-B),gnr_pdisk_wcache_enabled(rg-sno-190153-A/e1s01, rg-sno-190153-A/e1s02, rg-sno-190153-A/e1s03)

Fix

Show the WCE states of all the drives:

[root@sno-190153-A 4k]# for disk in $(lsscsi |grep -e "disk.*SEAGATE" | awk '{print $6}'); do sdparm -g WCE $disk; done ;
    /dev/sdb: SEAGATE   ST16000NM004J     E002
WCE           1  [cha: y, def:  1, sav:  1]
    /dev/sdc: SEAGATE   ST16000NM004J     E002
WCE           1  [cha: y, def:  1, sav:  1]
    /dev/sdd: SEAGATE   ST16000NM004J     E002
WCE           1  [cha: y, def:  1, sav:  1]
    /dev/sde: SEAGATE   ST16000NM004J     E002
WCE           1  [cha: y, def:  1, sav:  1]
    /dev/sdf: SEAGATE   ST16000NM004J     E002
WCE           1  [cha: y, def:  1, sav:  1]
    /dev/sdg: SEAGATE   ST16000NM004J     E002

Set WCE to a drive:

[root@sno-190153-A ~]# sdparm -s WCE=0 --save /dev/sdb

Set WCE to all the drives:

# for disk in $(lsscsi |grep -e "disk.*SEAGATE" | awk '{print $6}'); do sdparm -s WCE=0 --save $disk; done ;

Must restart GPFS of A and maybe also B to clear the state, this can be done one at a time:

mmshutdown
mmstartup