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, 4 Mar 2008, Florian Haas wrote: > On Tuesday 04 March 2008 15:39:34 drbd at bobich.net wrote: >>>> DRBD doesn't know and doesn't care if your RAID stripe is degraded or >>>> optimal. That's a job for something else. The underlying partition for >>>> DRBD is either there or isn't. There is nothing inbetween. If you want >>>> to fail over when the local RAID stripe is degraded, then you'll have to >>>> put in some external monitoring, possibly integrated with heartbeat. So >>>> yes, probably a Linux-HA question. >>> >>> Um, not quite. >>> http://www.drbd.org/users-guide/s-handling-disk-errors.html >> >> This is about handling disk errors. I was talking about whether the >> underlying RAID stripe is degraded or optimal. > > And would you mind enlightening me how a RAID-0 set (whose use you suggested > earlier) ever becomes "degraded" -- rather than faulty and issuing an I/O > error -- when losing a disk? > > Last time I checked, RAID-0 wasn't redundant. Hence why I said that it'd either be there or it wouldn't with nothing inbetween. There's cross-talk here between using redundant and non-redundant underlying RAID, as both were discussed. The original question was whether DRBD or something else handles the fail-over under conditions of a degraded redundant stripe. Gordan