From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <794f77b6ad37c299fe74831e40e5225b@swtch.com>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] fossil+venti continuous disk activity
From: "Russ Cox" <rsc@swtch.com>
Date: Thu, 23 Mar 2006 12:38:17 -0500
In-Reply-To: <db8eee8a5c7c1d948c84de650d285b50@terzarima.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Topicbox-Message-UUID: 1eaca16e-ead1-11e9-9d60-3106f5b1d025

sorry, i should have said more.

locking up for 10 minutes straight is not common behavior,
but in the scenario you described, it's to be expected.
fossil was in the middle of archiving its first snapshot
to disk.  let's call that 250MB.  the distributed venti is
robust but quite slow -- four long disk seeks per
write if you are running index and arenas on the same disk.
on a single disk i expect around 1-3GB/hour.  so 10 minutes
for that 250MB is entirely within the ballpark.

as forsyth noted, during archive snapshots fossil usually
does respond to requests, but you explicitly told it not to
by running fshalt.  it didn't finish halting until the archive
was done.  i suppose halt could stop the archiver and
let it pick up again later.  but it doesn't.

russ