Dear list,

I have a setup with a failover HAST + ZFS cluster (CARP and auto switch
scripts were deactivated long ago because of too much splitbrains).

I didn't do the setup, and I do not have an extensive experience on HAST.

I now have a drive failing (predictive failure) on the primary server
that I want to change. I'd like to avoid to switch servers (mainly
because zpool export/import is long and so there is an impacting service
downtime).

If I remove the disk on the primary host (or if it fails by itself), the
secondary will be used to serve data. I'm so asking myself if I can do a
replication from the secondary to the new primary.

I know this will degrades performances, but I prefer that to a downtime.

I saw on the table of https://wiki.freebsd.org/HAST (How does
synchronization work?)
that it will result to the "?! Full sync from secondary" synchronization
type, so, I hope it is possible.

I want to be sure that it won't result in a split brain situation.

This is the procedure what I was thinking to follow:

# remove the drive from raid controler:
mfiutil fail 0
# change the disk, then recreate jbod for it
mfiutil create -u0 create jbod E1:S0
# init hast device
hastctl init mfid1
# set hast device as primary
hastctl role primary mfid1

Is this the right thing to do ?
Should I directly set the hast device to primary ?
Will the replication start the right way (from secondary to primary) ?

Best regards,
-- 
Pablo Joubert
EMBLGR IT Team
_______________________________________________
[email protected] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-cluster
To unsubscribe, send any mail to "[email protected]"

Reply via email to