Date: Tue, 26 Jun 2007 16:54:05 +0200 From: Fluffles <etc@fluffles.net> To: Ivan Voras <ivoras@fer.hr> Cc: freebsd-geom@freebsd.org Subject: Re: gjournal performance issues Message-ID: <4681288D.20804@fluffles.net> In-Reply-To: <f5qgrf$25r$1@sea.gmane.org> References: <4680438E.9030407@fluffles.net> <f5qgrf$25r$1@sea.gmane.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Ivan Voras wrote: > Fluffles wrote: > >> First, my setup: >> 8 disks in RAID5 using geom_raid5, gjournal on top where both the >> journal (1GB) and the data is stored on the same consumer. Since >> gjournal uses both metadata and file/dir for journaling, this means >> that, theoretically, the write speed of sequential operations is >> doubled. Unfortunately, it appears to have crashed. > > Just to rule out geom_raid5, can you repeat your tests with geom_raid3? Sounds logical. But: RAID3 does not have any 2-phase writes, all writes are 1-phase, involving all disks. Thus RAID3 would be less affected by non-contigious writes. I will test with debuglevel or with geom_nop to see the I/O that geom_journal produces, and to see if this I/O is contigious or not. I hope to know more in a day or so. Thanks for your input Ivan and Pawel! More testing is needed... - Veronica
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4681288D.20804>