CEPH filesystem development
 help / color / mirror / Atom feed
From: Alex Elder <elder@ieee.org>
To: Ilya Dryomov <idryomov@gmail.com>
Cc: Ceph Development <ceph-devel@vger.kernel.org>,
	Varada Kari <Varada.Kari@sandisk.com>
Subject: Re: [PATCH 1/4] libceph: don't bail early from try_read() when skipping a message
Date: Sat, 20 Feb 2016 14:38:47 -0600	[thread overview]
Message-ID: <56C8CED6.7080701@ieee.org> (raw)
In-Reply-To: <CAOi1vP8V7Ni6G53Y-xhmuJihd=dmM6Fg6960CQZmLt451y5V5A@mail.gmail.com>

On 02/20/2016 02:21 PM, Ilya Dryomov wrote:
>> > So...  There are two places in try_read_message() where
>> > incoming messages should be discarded, and they represent:
>> > receipt of a duplicate (re-sent) message; receipt of an
>> > unexpected message type; receipt of a response message
>> > that was not expected; or receipt of a response that was
>> > too large for the buffer that had been allocated to hold it.
> Just a note: it's less about the preallocated buffer being too small
> part (in theory we could allocate everything in alloc_msg(), although
> that would be stupid) and more about the "no outstanding request is
> waiting for that response" part.  Stray and/or duplicate replies from
> the OSDs are expected in various cases.

Agreed.  That one case (buffer too large) is anomalous.  But
the code *does* say "skip the message" in that case, which
is why I mentioned it.

Like I said before, when it's been a while since I've reviewed
code I need to go immerse myself in it for a little while.  And
while I'm at it, I figure I might as well explain it all for the
benefit of other people.  This is especially true for these tiny
patches--sometimes the smaller the patch, the bigger the context
needed to understand why it fixes a problem.

					-Alex


  reply	other threads:[~2016-02-20 20:38 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-02-20 16:45 [PATCH 0/4] libceph: message skipping fixes Ilya Dryomov
2016-02-20 16:45 ` [PATCH 1/4] libceph: don't bail early from try_read() when skipping a message Ilya Dryomov
2016-02-20 18:28   ` Alex Elder
2016-02-20 20:21     ` Ilya Dryomov
2016-02-20 20:38       ` Alex Elder [this message]
2016-02-20 16:45 ` [PATCH 2/4] libceph: use the right footer size " Ilya Dryomov
2016-02-20 18:44   ` Alex Elder
2016-02-20 20:05     ` Ilya Dryomov
2016-02-20 20:15       ` Alex Elder
2016-02-21 14:31       ` Ilya Dryomov
2016-02-21 14:35         ` [PATCH 4/4] libceph: use sizeof_footer() more Ilya Dryomov
2016-02-21 15:08           ` Alex Elder
2016-02-20 16:45 ` [PATCH 3/4] libceph: don't spam dmesg with stray reply warnings Ilya Dryomov
2016-02-20 18:46   ` Alex Elder
2016-02-20 16:45 ` [PATCH 4/4] libceph: use sizeof_footer() in ceph_msg_revoke() Ilya Dryomov
2016-02-20 18:47   ` Alex Elder
2016-02-20 19:51     ` Ilya Dryomov
2016-02-20 19:56   ` Alex Elder

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=56C8CED6.7080701@ieee.org \
    --to=elder@ieee.org \
    --cc=Varada.Kari@sandisk.com \
    --cc=ceph-devel@vger.kernel.org \
    --cc=idryomov@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox