Is this still an issue? I'm seeing something very much like this: a loader process that started out running quickly has now slowed dramatically, with 75% of the cpu being given to PageStore.checkpoint() while fetching blocks of 100 ids from a sequence.
I'm using 1.3.171. On Thursday, March 11, 2010 9:34:17 PM UTC-8, [email protected] wrote: > > Thomas, > > That's excellent detective work; I was despairing of replicating it. > > I'll move the log size up significantly and let you know if that let's > me get to the current version. > > Thanks a lot! > > Chris > > On Mar 11, 3:16 pm, Thomas Mueller <[email protected]> > wrote: > > Hi, > > > > I can now reproduce the problem. The database writes many checkpoints > > unnecessarily, which slows down the operation. It does that because > > the the log "file" (no it's no longer a file, it's a segment) is too > > large. If there is an open transaction, it can't delete the old log > > segment however, so that it will create a new segment for each 32 (by > > default) sequences. I will fix that in the next release. > > > > A workaround is to use a larger max_log_size or smaller transactions. > > > > Regards, > > Thomas > > -- You received this message because you are subscribed to the Google Groups "H2 Database" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/h2-database. For more options, visit https://groups.google.com/groups/opt_out.
