From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Alan D. Brunelle" Date: Fri, 18 Jul 2008 13:24:34 +0000 Subject: Re: [Patch] blkiomon: repost with some fixes and improvements Message-Id: <48809992.20104@hp.com> List-Id: References: <1216377086.8497.4.camel@kitka.ibm.com> In-Reply-To: <1216377086.8497.4.camel@kitka.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable To: linux-btrace@vger.kernel.org T=F6r=F6k Edwin wrote: > On 2008-07-18 13:31, Martin Peschke wrote: >> recent changes: >> - added man page >> - beautified human-readable output >> - fixed an x86 compile error caused by incomplete endianess handling >> - fixed some x86 __u64 vs. unsigned long compiler warnings >> - fixed checking of a command line argument >> >> >> blkiomon periodicaly generates per devive request size and request laten= cy >> statistics from blktrace data. It provides histograms as well as data th= at >> =20 >=20 > Does it also measure latency caused by the request queues being full? > (which happens around here): > blk-core.c:get_request: > /* > * The queue is full and the allocating > * process is not a "batcher", and not > * exempted by the IO scheduler > */ > goto out; >=20 > That would be very useful to trace some high latencies I am seeing, see > discussion here: > http://lkml.org/lkml/2008/7/12/104 >=20 > Best regards, > --Edwin Hi Edwin - It would probably be pretty hard to do so - Martin is mostly interested in issue & complete information. The stuff you needs comes from sleep requests. The btt tool is static, but does output some data concerning the amount of time spent "sleeping" - waiting for a request to be made available. (But only for the whole duration of the run, you're looking for an instantaneous output.) I've also looked at seekwatcher - but it too just looks at the issue & complete stuff, but I will see if there is some way of handling sleep request stuff "live". Alan