(Illustration by Gaich Muramatsu)
On Sat, Mar 11, 2006 at 12:03:46AM +0200, Denis Chapligin wrote: > tau:~# rvmutl /vice/LOG > * status > space used by records: 772 > space available: 20970748 > status area size: 1536 > total log size: 20973056 > > first record number: 541352 > last record number: 541352 There are only 772 bytes in the log, it looks like the server just performed a truncation, which means that pretty much everything already got committed to the data segment. It then tried to log one more operation but somehow died (power failure?) before it had a chance to complete the operation. Not entirely sure why RVM is unable to recover by rolling back this failed operation, I'm pretty sure the default behaviour is to only try to apply complete records and abort/ignore any partially logged operation because it's consistency is suspect. You can make a copy of the current RVM log and data files if you want, then reinitialize the log as follows, rvmutl * i /vice/LOG 20M * q This will reinitialize just the log and leave the data alone. If the server is then restarted it won't have anything to recover and use whatever was previously committed to RVM data. JanReceived on 2006-03-11 23:54:36