linux-bcache.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Rolf Fokkens <rolf@rolffokkens.nl>
To: linux-bcache@vger.kernel.org
Subject: bcache_writebac task 100% CPU
Date: Sun, 18 May 2014 10:23:30 +0200	[thread overview]
Message-ID: <53786E02.8060300@rolffokkens.nl> (raw)

Sometimes bcache_writebac task consumes 100% CPU during normal operation, but after a while (like a minute) this drops to normal.

It can be triggered in a reproducable fashion on my system by:

echo 0 > /sys/block/bcache0/bcache/writeback_percent

After this bcache_writebac starts to burn CPU cycles, and it never seems to stop doing so. changing the writeback_percent value back to 25 doesn't change anything.

Apart from a busy CPU there appear to be no problem (e.g. no "task bcache_writebac:286 blocked for more than 120 seconds" messages).

I filed this issue also on: https://bugzilla.kernel.org/show_bug.cgi?id=76391

[rolf.fokkens@home07 ~]$ bcache-status  -s
--- bcache ---
UUID                        4892b3d5-fd49-4d71-adb6-d7b0a799e913
Block Size                  512 B
Bucket Size                 512.00 KiB
Congested?                  False
Read Congestion             0.0ms
Write Congestion            0.0ms
Total Cache Size            30 GiB
Total Cache Used            2 GiB	(6%)
Total Cache Unused          28 GiB	(94%)
Evictable Cache             30 GiB	(100%)
Replacement Policy          [lru] fifo random
Cache Mode                  writethrough [writeback] writearound none
Total Hits                  31727	(99%)
Total Misses                76
Total Bypass Hits           3569	(100%)
Total Bypass Misses         0
Total Bypassed              27.5 MiB
--- Backing Device ---
   Device File               /dev/md2 (9:2)
   bcache Device File        /dev/bcache0 (252:0)
   Size                      140 GiB
   Cache Mode                writethrough [writeback] writearound none
   Readahead                 0
   Sequential Cutoff         0 B
   Merge sequential?         False
   State                     dirty
   Writeback?                True
   Dirty Data                6.1 MiB
   Total Hits                31727	(99%)
   Total Misses              76
   Total Bypass Hits         3569	(100%)
   Total Bypass Misses       0
   Total Bypassed            27.5 MiB
--- Cache Device ---
   Device File               /dev/sdd1 (8:49)
   Size                      30 GiB
   Block Size                512 B
   Bucket Size               512.00 KiB
   Replacement Policy        [lru] fifo random
   Discard?                  True
   I/O Errors                0
   Metadata Written          906.00 KiB
   Data Written              52.3 MiB
   Buckets                   61440
   Cache Used                2 GiB	(6%)
   Cache Unused              28 GiB	(94%)
[rolf.fokkens@home07 ~]$

             reply	other threads:[~2014-05-18  8:23 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-05-18  8:23 Rolf Fokkens [this message]
2014-05-19 16:55 ` bcache_writebac task 100% CPU Kent Overstreet
2014-05-19 18:01   ` Rolf Fokkens
2014-05-20 19:38   ` Rolf Fokkens

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=53786E02.8060300@rolffokkens.nl \
    --to=rolf@rolffokkens.nl \
    --cc=linux-bcache@vger.kernel.org \
    /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;
as well as URLs for NNTP newsgroup(s).