FreeBSD 8.1 and HAST

[Mikolaj Golub]

On Thu, 17 Jun 2010 12:45:52 +0900 hiroshi at soupacific.com wrote:

 h> Hi folks !

 h> Couple more question now ! Since my test environment is 1T bytes hdd,
 h> so it took long long time to finish first synchronize dirty bytes.

 h> When manual start HAST on each node, then it will start synchronize
 h> dirty bytes. During synchronization,
 h> set rc.conf

 h> hast_enable="YES"
 h> ucarp_enable="YES"

 h> After reboot,
 h> In ucarp_up.sh, zpool import -f poolname shows error as

 h> ZFS pool import from resource failed: cannot import 'hasthoem': pool
 h> is formatted using a newer ZFS version.

 h> After completion of synchronization, then switch slave as
 h> master. using /etc/vip_up.sh

 h> Same ZFS pool import error message appears

 h> Also
 h> hastctl status shows
 h> status: degaraded

 h> Mr. Mikolaj Golub ! Did you have this problem before ?

I didn't. But as I wrote earlier I worked with HAST on CURRENT. So I just
checked if there is some issue with ZFS pools on HAST 8-STABLE. I created
400Mb md device, configured HAST to use it and created zpool. Manual switching
to failover:

host1:
zpool export -f storage
hastctl role secondary storage

host2:
hastctl role primary all
zpool import -f storage

There were no any issues.

Just a note, as I don't have many 8-STABLE boxes this HAST has been created
between my host in the office and my laptop at home. Very nice to see that it
works via WAN :-)

So, can you try recreating ZFS pool to see if "zpool import" issue has gone?

Please provide here hastctl status on both nodes and hastctl.cfg. Then we
might ask you about logs :-).

I would recommend playing at first without ucarp -- switching to failover
manully. In this way you can be sure that HAST works and also be more familar
how it works. After this you can try ucarp. I have not used ucarp myself -- I
have been using our own application for failure detection and initiating
switching to failover.

-- 
Mikolaj Golub