From mboxrd@z Thu Jan 1 00:00:00 1970 From: Phil Turmel Subject: Re: RAID6 reshape stalls immediately Date: Mon, 9 Nov 2015 17:26:31 -0500 Message-ID: <56411D97.7020705@turmel.org> References: <84a8qtgwqk.wl-peter.chubb@nicta.com.au> <563CAFC1.2080708@gmail.com> <84611dw2cw.wl-peter.chubb@nicta.com.au> <56409F4E.5000508@gmail.com> <5640A1ED.3010205@turmel.org> <84lha6u90k.wl-peter.chubb@nicta.com.au> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <84lha6u90k.wl-peter.chubb@nicta.com.au> Sender: linux-raid-owner@vger.kernel.org To: Peter Chubb Cc: Anugraha Sinha , linux-raid@vger.kernel.org List-Id: linux-raid.ids On 11/09/2015 05:02 PM, Peter Chubb wrote: > I just did > mdadm --grow --continue /dev/md0 > > and the reshape started up again. I didn't realise that mdadm had to > keep going to do the reshape -- it must have died when I logged off > last time. So many other operations are handled entirely in the > kernel... mdadm spawns a copy of mdmon to perform these tasks. If the process hierarchy is strictly enforced, and kills off spawned processes when the highest parent dies, that's a problem for mdadm. I recall some discussions about this with systemd and/or cgroups. You might want to make sure all of your utilities have the latest fixes. Phil