mbbackup taking too long, potentially locking itself
started an mbbackup on a 3 node cluster - 1.7.1, each holding 8 gb of data.
after ~20 hours, none of the backups are complete.
some seem to be progressing (by watching the file sizes), but another seems stuck (no file size increases.)
on that machine in particular, I see this line in TOP:
23212 root 20 0 24840 6412 988 R 100 0.0 757:45.23 sqlite3
There are quite a few factors that come into play here to determine the speed at which sqlite can backup it's files:
-Other I/O load on the system (including Membase writing to the disk). In some cases, it makes sense to create the backup on another drive to separate the live IO from the backup IO
-Ongoing writing to sqlite. If Membase is constantly writing, it will affect the speed at which sqlite can take a backup
-Internal fragmentation of the file.
Hope that helps...500mb an hour seems a bit slow but not unheard of.
Perry
Okay. Two of them have finished. The third is perhaps making progress, now.
Less than 500MB an hour?