Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Hi, My understanding is that "verify" has a source and a target - and then when you resync, blocks will flow from the source to the target, so if the source is corrupt you'll replicate that corruption. Anyone please correct me if I'm wrong. Lionel Sausin. Le -10/01/-28163 20:59, Dominik Klein a écrit : > Anyone? > > Thanks > Dominik > > On 06/28/2011 08:29 AM, Dominik Klein wrote: >> Hi >> >> while recovering from a failure that impacted a node's power and its >> raid controller's battery (ouch), I came across a question regarding >> drbdadm verify I could not find the answer to. Maybe you guys can shed >> some light. >> >> In my situation, drbd thinks it did write data that might not have made >> it to disk (read: simultaneous power and raid ctrl battery failure). I >> can see that the root fs misses data about 1 minute of logs, so I assume >> something like this for the drbd partitions as well. >> >> So, say there is a cronjob on the node that had the outage which runs >> drbdadm verify every day. Would blocks that changed on node2 (no outages >> there) in the meantime be overwritten with what is empty/bad on node1? >> >> I know it wouldn't be overwritten right away, just marked oos, but what >> would happen on the next disconnect/connect? >> >> Regards >> Dominik >