Date: Mon, 13 May 1996 17:34:48 -0400 (EDT) From: Brian Tao <taob@io.org> To: "Amancio Hasty Jr." <hasty@rah.star-gate.com> Cc: Joe Greco <jgreco@brasil.moneng.mei.com>, FREEBSD-CHAT-L <freebsd-chat@freebsd.org> Subject: Re: 3 terabytes on one server? (was Re: more than 32 scsi disks on a single machine ?) Message-ID: <Pine.NEB.3.92.960513172246.14554X-100000@zot.io.org> In-Reply-To: <199605132116.OAA00293@rah.star-gate.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 13 May 1996, Amancio Hasty Jr. wrote:
>
> OTOH, on my FreeBSD box it takes about 22.5 seconds per GIG which means
> that if your news server was on a FreeBSD it will take about 6.75 seconds.
You mean 6.75 minutes (not seconds). :) When the news server
crashes and reboots, there is almost exactly 10 minutes missing in the
syslog. There are also over 2.5 million files, which probably extends
the time of an fsck. You'll note that the kernel panic and subsequent
boot messages aren't recorded in the syslog. :(
May 11 11:18:14 news1 nnrpd[16933]: myrtle.net3.io.org group io.support.system 1
May 11 11:18:17 news1 last message repeated 2 times
May 11 11:18:17 news1 nnrpd[16886]: dbrin.net6a.io.org group alt.binaries.pictures.erotica 1
May 11 11:28:13 news1 /kernel: e segment = base 0x0, limit 0xfffff, type 0x1b
May 11 11:28:13 news1 /kernel: = DPL 0, pres 1, def32 1, gran 1
May 11 11:28:13 news1 /kernel: processor eflags = interru
May 11 11:28:14 news1 lpd[103]: restarted
May 11 11:28:17 news1 innd: ME descriptors 2048
May 11 11:28:17 news1 innd: ME outgoing 2035
> Now if you parallelized your fsck like someone posted here then it should take
> less than 2 minutes to fsck 18 gigs 8)
Well, I fiddled around a bit with options in my /etc/fstab, and I
still can't get fsck to run in parallel over several drives. I'm
using this on my FTP server now:
/dev/sd0a / ufs rw 1 1
/dev/sd0s1e /usr ufs rw 1 1
/dev/sd0s1f /var ufs rw 1 1
/dev/sd0s1g /usr/local ufs rw 1 1
/dev/sd0s1h /tmp ufs rw 1 1
/dev/sd1s1e /usr/httpd ufs rw 1 2
/dev/sd1s1f /var/spool/ftp/pub ufs rw,userquota=/var/quotas/pub 1 2
/dev/sd2s1e /var/spool/ftp/.1 ufs rw 1 2
/dev/sd3s1e /var/spool/ftp/.2 ufs rw 1 2
Whether I use '2' or '1' for the non-root disks, it takes 5:50 to
fsck the four drives. sd1 to sd3 are 4GB drives, so parallelizing an
fsck should have a noticeable difference. I'm invoking it as "fsck -y"
as well as "fsck -y -l 10". What am I missing?
> Oh, I have 54000 rpm disks ...
Amazing! Send a bunch over to Satoshi so we can test FreeBSD on
extremely large, extremely fast filesystems. :)
--
Brian Tao (BT300, taob@io.org, taob@ican.net)
Systems and Network Administrator, Internet Canada Corp.
"Though this be madness, yet there is method in't"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.NEB.3.92.960513172246.14554X-100000>