From mboxrd@z Thu Jan 1 00:00:00 1970 From: Konrad Rzeszutek Subject: Re: [PATCH multipath-tools-0.4.7.rhel5.13 1/1] multipathd excessive path checker logging Date: Fri, 26 Jun 2009 21:39:21 -0400 Message-ID: <20090627013921.GA2580@mars.virtualiron.com> References: <20090626205057.GL3172@ether.msp.redhat.com> Reply-To: device-mapper development Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20090626205057.GL3172@ether.msp.redhat.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: dm-devel-bounces@redhat.com Errors-To: dm-devel-bounces@redhat.com To: device-mapper development List-Id: dm-devel.ids On Fri, Jun 26, 2009 at 03:50:57PM -0500, Benjamin Marzinski wrote: > On Thu, Jun 25, 2009 at 01:50:51PM -0400, Charlie Brady wrote: > > > > multipathd logs a path checker message whenever state changes. But it also > > logs every non-changed message when verbosity is >=4 and newstate is > > PATH_UP or PATH_GHOST, and every non-changed message when verbosity is >= 2 > > and newstate is PATH_DOWN. > > > > I believe that the message should only be logged, once, when state changes. > > I disagree. If you look at the stuff that gets logged when verbosity >= > 4, you'll see that it's a whole bunch of stuff that normal users would > never care about, but may be useful for debugging. This clearly fits. > > I also think that having a path down is something that shouldn't just > blip by once in the logs. It's something that often requires the > sysadmin to intervene. I could see limitting it to, say, only happen > on every tenth pass through the checker loop, so that these messages > don't clog up the logs so bad. But since there is the option to set the > verbosity to 1 to avoid this, I think that printing the path down > messages repeatedly is useful. Hmm.. What if you have 100 LUNs with four paths, meaning you get 400 messages saying: "Down! Down! Pants on fire!"