From mboxrd@z Thu Jan 1 00:00:00 1970 From: Harald Hoyer Subject: Re: Does dracut need start watchdog? Date: Fri, 09 Aug 2013 09:20:53 +0200 Message-ID: <52049855.6050308@redhat.com> References: <20130805101929.GA26184@dhcp-16-252.nay.redhat.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20130805101929.GA26184-je1gSBvt1Td8jiCoVU1BWR/sF2h8X+2i0E9HWUfgJXw@public.gmane.org> Sender: initramfs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: Content-Type: text/plain; charset="us-ascii" To: Baoquan He Cc: initramfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org On 08/05/2013 12:19 PM, Baoquan He wrote: > Hi Harald, > > Since some drivers do the soft kicking. That is if BIOS has enabled > watchdog, then kernel driver kicks the hardware till user space watchdog > daemon comes along. This can help sove the problem in kdump which kdump > kernel reboot before vmcore saving is finished. That problem is > triggered because watchdog started in 1st kernel and timeout because of > stopping being kicked. > > I noticed in 04watchdog the watchdog.sh will start watchdog if > /dev/watchdog exists. I am just wondering if it's needed to start > watchdog by accessing /dev/watchdog. If starting watchdog is not needed, > I want to make some code change of watchdog.sh. > > What's your suggestion? > > Baoquan > Thanks a lot Basically the 04watchdog module was used only in the testsuite. You might consider to use the systemd watchdog feature instead. http://0pointer.de/blog/projects/watchdog.html or, if 0pointer is down again: http://webcache.googleusercontent.com/search?q=cache:jYfJU_fFZy8J:0pointer.de/blog/projects/watchdog.html