Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Sunday 06 June 2010 07:44:41 Olivier Le Cam wrote: > Hi - > > I have a simple DRBD+NFS Active/Passive system with Heartbeat (broadcast > and crossover serial cable, and "auto_failback on"). > > When node1 (the Primary node) is off-lined (shutdowned), node2 become > Primary as expected. > > When the node1 is powered on again, DRBD perfectly resyncs the data back > from node2 to node1 but for some reason node1 quickly regain the Primary > state although /proc/drbd shows that data are still Inconsistent on it. > > Here are some screenshots taken on node2: > > 1) node1 off line, node2 running as Primary: > > 0: cs:WFConnection st:Primary/Unknown ds:UpToDate/DUnknown C r--- > ns:631836 nr:14915164 dw:15759500 dr:1362578 al:2532 bm:3766 lo:0 > pe:0 ua:0 ap:0 > resync: used:0/61 hits:0 misses:0 starving:0 dirty:0 changed:0 > act_log: used:0/257 hits:208552 misses:2721 starving:0 > dirty:189 changed:2532 > > > > 2) node1 back on line, resyncing is in progress > > 0: cs:SyncSource st:Primary/Secondary ds:UpToDate/Inconsistent C r--- > ns:672732 nr:14915164 dw:15759500 dr:1406078 al:2532 bm:3963 lo:0 > pe:17 ua:82 ap:0 > [===>................] sync'ed: 21.2% (170444/210800)K > finish: 0:00:04 speed: 40,356 (40,356) K/sec > resync: used:2/61 hits:3465 misses:203 starving:0 dirty:0 > changed:203 > act_log: used:0/257 hits:208552 misses:2721 starving:0 > dirty:189 changed:2532 > > > 3) node1 gained Primary state although it is still Inconsistent > > 0: cs:SyncSource st:Secondary/Primary ds:UpToDate/Inconsistent C r--- > ns:780112 nr:14915164 dw:15761292 dr:1513566 al:2578 bm:4351 lo:0 > pe:267 ua:467 ap:0 > [============>.......] sync'ed: 69.3% (67088/210800)K > finish: 0:00:02 speed: 28,740 (28,740) K/sec > resync: used:24/61 hits:13396 misses:433 starving:0 dirty:0 > changed:433 > act_log: used:0/257 hits:208954 misses:2767 starving:0 > dirty:189 changed:2578 > > > Is it the normal behaviour? My guess is that node1 could not gain again > the Primary role before its data is UpToDate... > > Thanks in anticipation for your comments. > > Best regards, > Sounds like a Heartbeat issue. Heartbeat should not make the node primary before the sync has finished. What resource agent did you use? B.