Date: Thu, 14 Dec 2006 09:17:02 -0700 From: Scott Long <scottl@samsco.org> To: Dmitry Pryanishnikov <dmitry@atlantis.dp.ua> Cc: Jan Mikkelsen <janm@transactionware.com>, freebsd-stable@freebsd.org Subject: Re: g_vfs_done() failures on 6.2-RC1 Message-ID: <458178FE.7050609@samsco.org> In-Reply-To: <20061214144023.K14515@atlantis.atlantis.dp.ua> References: <00a601c71e7f$ed63f7a0$3301a8c0@janmxp> <457FAAFD.1080707@samsco.org> <00d401c71e8d$fb60de00$3301a8c0@janmxp> <20061214144023.K14515@atlantis.atlantis.dp.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
Dmitry Pryanishnikov wrote: > > Hello! > > On Wed, 13 Dec 2006, Jan Mikkelsen wrote: >> I mentioned that I have seen similar problems on machines with >> ataraid, like this: >> >> DOH! ata_alloc_composite failed! (x5) >> FAILURE - out of memory in ata_raid_init_request (x6) >> g_vfs_done():ar0s3f[WRITE(offset=113324673024, length=2048)]error = 5 > > These errors are caused by the kernel memory (specifically, > "ata_composite_zone" UMA zone) depletion. To get this knowlege, one > should Read The F[ine] Sources, as ata(4)/ataraid(4) lack the > DIAGNOSTICS section: > > ata-all.h:#define ata_alloc_composite() uma_zalloc(ata_composite_zone, > M_NOWAIT | M_ZERO) > > Sincerely, Dmitry Not being able to perform an I/O due to lack of memory is a very bad thing, since the way that memory is freed up is by doing I/O. Scott
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?458178FE.7050609>