Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
hi:
i am running drbd 8.4.5 (from epel) and pacemaker under
scientific linux 7.0.
my drbd host have 3 nic teamd (with roundrobin protocol) to
connect peer host. yesterday maybe something wrong with one nic, so
about 1/3 packets are lost when I ping each other. drbd link is broken
under the situation. pacemaker is fine since it has another network
link besides the teamd connection.
when the link is broken, drbd block IO and called
"/usr/lib/drbd/crm-fence-peer.sh". but somehow it didn't release the
block after putting the constraint. I need the hosts to back on line,
so I just reset the host and resync drbd. I didn't think much when I
met the situation. I was worried about data corruption. but I think I
can do better next time, maybe I should release the block by hand?
will "drbdadm resume-io" do the trick?
the log is at link below: http://pastebin.com/RNZJ5YcE
I don't know why drbd won't release block? maybe the link is not
totally dead?
thanks a lot for advice!