order

[lelunicu]

Hi,

in https://sort.veritas.com/public/documents/sf/5.0MP3/linux/html/vcs_notes/ch01s12s01s05.htm is written repeat steps 3 to 7 on each of those nodes.

I think better is 3 to 8.

I mean after i apply patch on first node and reboot i switch back the service on node 1-then i take node 2.

True?

[frankgfan]

step 8 maybe not even needed as restarting HAD may result in the service groups be onlined on the "original node" designated (this is decided by your VCS configuration as well as VCS version).  Also this kind of OS upgrade is normally performed during a maintenance window meaning all the production will be down for some times, switch the service groups as the end is fine

[lelunicu]

if i like to patch with no downtime then

hastop -all -force

patch every node 

and hastart on dode one by one.

true?

[frankgfan]

before running hastop -force -all,, run the commands below

hastatus -sum         <<< make sure no resource/service groups in transient state  (onlining/offlining/failing)

hagrp -freeze -persistent SG_name    <<< freeeze all the service group to prevent unexpected on-/offline during maintenance

hastop -all -force

follow vendor's procedures to patch up OS   (make sure Veritas version is compatible with the new OS patch)

start VCS on each node

unfreeze all the SGs frozen