I am using a separate private ethernet for the heartbeat, I shutdown
one (P) server the other (S) takes the primary role and goes in
standalone mode Primary/Unknown. Till here it is OK.<br>
<br>
Now when the main (P) server comes back it takes over as Primary/Unknown state:standalone.<br>
And the other one goes to Secondary/Unknown state:standalone.<br>
<br>
They do not connect with each other automatically? shouldn't they?<br>
--<br>
Usman<br><br><div><span class="gmail_quote">On 9/15/06, <b class="gmail_sendername">Tim Jackson</b> <<a href="mailto:lists@timj.co.uk">lists@timj.co.uk</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Christophe Zwecker wrote:<br><br>> node1 is primary with mounted fs<br>> node2 is secondary<br>><br>> nod1 goes down (only network failure),<br><br>"only" network failure? Which network? In many cases, a network failure
<br>alone is worse than one box completely failing, because it can cause<br>"split brain" if you're not careful.<br><br>What connections do you have for Heartbeat to use? (A serial heartbeat<br>is always a good idea if you can have it). As many redundant paths as
<br>possible is good. (typical might be 3: replication (crossover) network<br>between the DRBD machines, "normal" network and serial heartbeat)<br><br>> heartbeat unmounts the drbd fs on<br>> node1. node 2 takes over and mounts the drbd volume.
<br><br>And what happens to node1 here? Are you sure that Heartbeat stops the<br>DRBD services? My guess is that you have a single network connection for<br>both DRBD and Heartbeat, in which case DRBD will still be primary on node1.
<br><br>> node1 comes backup, mounts drbd volume and the change aint there because:<br>> Sep 15 13:47:03 mw-test-n2 kernel: drbd0: Current Primary shall become<br>> sync TARGET! Aborting to prevent data corruption.
<br><br>DRBD is doing the right thing here. Either your nodes weren't really<br>synchronised before the failure, or you had a split brain where DRBD was<br>primary on both machines.<br><br>This situation can only be resolved manually,
i.e. by a human telling<br>DRBD which machine has the latest data. (something like "drbdadm XXX<br>invalidate_remote --do-what-I-say" on the "good" machine)<br><br>Tim<br>_______________________________________________
<br>drbd-user mailing list<br><a href="mailto:drbd-user@lists.linbit.com">drbd-user@lists.linbit.com</a><br><a href="http://lists.linbit.com/mailman/listinfo/drbd-user">http://lists.linbit.com/mailman/listinfo/drbd-user</a>
<br></blockquote></div><br><br clear="all"><br>-- <br>Usman Ahmad Malik