From the timestamps, that's decaying way too fast. Looks like a bug.

On Thu, Jan 26, 2012 at 10:38 AM, Adam Berkan [off-list ref] wrote:

The /sys/block/bcache0/bcache/stats_total/bypassed should accumulate
(notice the stats_total), but stats_day only shows results with an
exponentially weighted average with a half life of 1 day.  If your
load varies during the day, the amount bypassed will swing up and down
accordingly.

Adam

On Thu, Jan 26, 2012 at 10:31 AM, David Rhodes Clymer
[off-list ref] wrote:

quoted

Although the doc (Documentation/bcache.txt) seems to indicate that
"bypassed" should contain cumulative IO for the stat period, I've
noticed that it does not consistently increase. In fact, it increases
and decreases quite a bit over a short period of time.

root@zapazoid:~# date
Thu Jan 26 13:23:34 EST 2012
root@zapazoid:~# cat /sys/block/bcache0/bcache/stats_day/bypassed
5.4M
root@zapazoid:~# cat /sys/block/bcache0/bcache/stats_day/bypassed
9.1M
root@zapazoid:~# cat /sys/block/bcache0/bcache/stats_day/bypassed
1008k
root@zapazoid:~# cat /sys/block/bcache0/bcache/stats_day/bypassed
2.4M
root@zapazoid:~# cat /sys/block/bcache0/bcache/stats_day/bypassed
5.1M
root@zapazoid:~# date
Thu Jan 26 13:24:44 EST 2012


This is while doing an rsync into a cached filesystem, which I would
imagine would entail a significant amount of sequential writes.

Am I just misunderstanding what this is supposed to be telling me?

-davidc
--
To unsubscribe from this list: send the line "unsubscribe linux-bcache" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe linux-bcache" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html