Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Tue, Mar 11, 2008 at 12:41:00PM -0400, Charles Riley wrote: > Hi, > > I have a drbd cluster which has experienced the following chain of > events (DRBD .7.24): > > All network interfaces on the primary node were restarted, to include > the DRBD interface. > Heartbeat at this point lost connectivity to it's ping nodes, and failed > over to the secondary node. > During the failover to the secondary node, the secondary node's DRBD was > set from Secondary/Unknown --> Primary Unknown and all services were > failed over. So far so good. > However, the secondary node's drbd was stuck in WFConnection. This went > unnoticed. > The system ran for a period of time, and then was failed back to the > primary node, causing us to jump back in time to the date of the initial > failover as far as the filesystem was concerned. > When it was noticed that the filesystem was out of date, the data on it > was restored from external sources. > Now I am brought in to the middle of this, and find that the primary > node (which now has good data) is 0: cs:StandAlone st:Primary/Unknown > ld:Consistent > The secondary is 0: cs:WFConnection st:Secondary/Unknown ld:Consistent > > My question is, what is going to happen when I reestablish communication > between primary and secondary? > Is drbd going to do the right thing and bring the secondary node's disk > up to date? > I'd just do a "drbdadm -- --do-what-I-say primary all" on the primary, > but I'm pretty sure that once the failover node's drbd starts > communicating again, a sync is going to start. > I just want to be prepared (and coordinate downtime) if I am going to > have to recover again from external sources. well, --do-what-I-say does not do what you mean. again. :) because it _is_ already primary, that would be a no-op. I suggest you do bad# cat /proc/drbd 0: cs:WFConnection st:Secondary/Unknown ld:Consistent bad# drbdadm disconnect all ; cat /proc/drbd 0: cs:StandAlone st:Secondary/Unknown ld:Consistent bad# drbdadm invalidate all ; cat /proc/drbd 0: cs:StandAlone st:Secondary/Unknown ld:Inconsistent bad# drbdadm adjust all ; cat /proc/drbd 0: cs:WFConnection st:Secondary/Unknown ld:Inconsistent good# cat /proc/drbd 0: cs:StandAlone st:Primary/Unknown ld:Consistent good# drbdadm adjust all ; sleep 15 ; cat /proc/drbd 0: cs:SyncSource st:Primary/Secondary ld:Consistent -- : Lars Ellenberg http://www.linbit.com : : DRBD/HA support and consulting sales at linbit.com : : LINBIT Information Technologies GmbH Tel +43-1-8178292-0 : : Vivenotgasse 48, A-1120 Vienna/Europe Fax +43-1-8178292-82 : __ please use the "List-Reply" function of your email client.