dlnkmgr view -drv
view dlm devices.dlmrmdev
deletes dlm devices. Must be done before deleting underlying hdisk devices.dlmpr -a
clears all persistent reservations
Uninstall:
installp -u DLManager.mpio.rte
installp -u Hitachi.hacmp.support
installp -u Hitachi.aix.support.rte
emcpreg -add some-key
to register powerpath software.lsattr -El fscsixx
, may want to seedynamic_tracking
andfast fail
are onlsattr -El device
check q depth (recommend 16) for all EMC or Powerpath devices seen inlsdev -Cc disk
no -p -o rfc1323=1
allows setting values greater than 4K; without this, send/recv sizes can be set higher, but will only use first 4K.tcpdump
puts interface into promiscuous mode
lssyscfg -r lpar -m msys --osrefresh
force refresh os information on HMClslock -e frame-name
list which HMC has a lockviosvrcmd -m msys -p vio-name -c errlog
run command on vio from HMCviosbr
dump VIOS config to a file. Use-view
to view as a report.print "cvai" | kdb | grep vscsi | grep -v read
show vio name and vhost from client lpar- LMB size must be same on all systems for LPM (use
lshwres
with-F mem_region_size
to view) NIMBUG
env variable - debug nimalog -f file -o
to view alog format files (e.g. nimlog)vmo -o nokilluid=uid_num
tell aix not to kill processes owned by this user when out of memoryrmcdomainstatus -s ctrmc
test rmc connection,I
= up,O
= down.pmcycles
show CPU speed
- 12 cores per chip
- 8 SMT threads
- S824 up to 2TB
- CAPI adapters <500 instructions (vs 20K) to get to SSD disk
- appears like slow memory rather than fast storage
- combine 795 & 770 & 780
- 16TB memory
- 19" rack
- mandatory midplane, with nodes (1 to 4) above and below
- no disk (SAS) in nodes; need IO drawer
- 32 cores per node; 4 node has 128 cores
If user A has sugroups = x,y,z then all users in groups x,y,z can su to user A
Force a primary SEA (currently in auto mode) to fail over to standby...
chdev -l <SEA device> -a ha_mode=standby
Bring it back...
chdev -dev <SEA device> -attr ha_mode=auto