From owner-freebsd-fs@freebsd.org Thu Jul 30 12:38:07 2015 Return-Path: Delivered-To: freebsd-fs@mailman.ysv.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by mailman.ysv.freebsd.org (Postfix) with ESMTP id C50869ADB5D for ; Thu, 30 Jul 2015 12:38:07 +0000 (UTC) (envelope-from lev@FreeBSD.org) Received: from onlyone.friendlyhosting.spb.ru (onlyone.friendlyhosting.spb.ru [46.4.40.135]) by mx1.freebsd.org (Postfix) with ESMTP id 8C63C15FD for ; Thu, 30 Jul 2015 12:38:07 +0000 (UTC) (envelope-from lev@FreeBSD.org) Received: from lion.home.serebryakov.spb.ru (unknown [IPv6:2001:470:923f:1:2924:7e01:7d9c:bbfe]) (Authenticated sender: lev@serebryakov.spb.ru) by onlyone.friendlyhosting.spb.ru (Postfix) with ESMTPSA id 82CC22D9F; Thu, 30 Jul 2015 15:38:04 +0300 (MSK) Date: Thu, 30 Jul 2015 15:37:58 +0300 From: Lev Serebryakov Reply-To: lev@FreeBSD.org Organization: FreeBSD X-Priority: 3 (Normal) Message-ID: <843537366.20150730153758@serebryakov.spb.ru> To: Steven Hartland CC: freebsd-fs@freebsd.org Subject: Re: ZFS on 10-STABLE r281159: programs, accessing ZFS pauses for minutes in state [*kmem arena] In-Reply-To: <55BA0F41.6070508@multiplay.co.uk> References: <164833736.20150730143008@serebryakov.spb.ru> <55BA0F41.6070508@multiplay.co.uk> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.20 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 30 Jul 2015 12:38:07 -0000 Hello Steven, Thursday, July 30, 2015, 2:49:21 PM, you wrote: > You don't need to do that as record set size is a min not a max, if you > don't force it large files will still be stored efficiently. Oh, I thought it is max after reading some blog-posts and forum threads (this one, for example: https://forums.freebsd.org/threads/1mb-recordsize-performance-recordsize-discussion.50414/ and this one https://blogs.oracle.com/roch/entry/tuning_zfs_recordsize and this https://www.joyent.com/blog/bruning-questions-zfs-record-size) Does it mean, that files could not occupy less than 128K with default setting? Anyway, I'll recreate pool. >> It looks FreeBSD contains bug about year ago which leads to this behavior, >> but mailing lists says, that it was fixed in r272221, 10 months ago. > When this happens what is the state of memory on the machine? Now I'm in single-user mode and have some problems to analyze massive output of "vmstat -z" :) top shows 4.5-5G of total ARC size and very low (~150Mb) free memory. > Top will give a good overview, while sysctl vm.stats.vm and vmstat -z > will provide some detail. > If you're seeing significant memory pressure, which could well be the > case with a mixed ZFS UFS system during this transfer (they use Now I have pure ZFS situation, with ZFS to ZFS transfer. Mixed scenario with UFS to ZFS transfer didn't show this behavior. > competing memory resource pools) then you could try limiting ARC via > vfs.zfs.arc_max I'll try to set it ot 3G... > You could also see if the patch on > https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=187594 help. I should try this, but for now I should migrate as quickly as possible & return my spare HDDs which are lent to me for short time... -- Best regards, Lev mailto:lev@FreeBSD.org