From: Kent Overstreet <koverstreet-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
To: Javier Marcet <jmarcet-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Cc: Linux BCache Mailing List
<linux-bcache-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>
Subject: Re: Issues dropping vm fs caches with kernel 3.5.2 + bcache master
Date: Mon, 20 Aug 2012 17:37:13 -0700 [thread overview]
Message-ID: <20120821003713.GB19615@google.com> (raw)
In-Reply-To: <CAAnFQG_rE9f61Rgo+xH6uNujLwNx_QFaF-OpH_=pYro-+99AFQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
On Tue, Aug 21, 2012 at 02:31:18AM +0200, Javier Marcet wrote:
> >> >> I've recently set up a software RAID5 + bcache.
> >> >>
> >> >> I'm using kernel 3.5.2 merged with bcache master branch. I'm quite
> >> >> happy with the performance but
> >> >> there is an important issue. I can't drop caches. Writing any number
> >> >> to /proc/sys/vm/drop_caches
> >> >> makes the system hung.
> >> >>
> >> >> Furthermore, so far I've been unable to curb bcache's memory
> >> >> consumption using the available
> >> >> sysctl knobs.
> >> >>
> >> >> Is this problem known?
> >> >
> >> > No I didn't know about that - thanks, I'll look into it.
> >
> > Just tried it - I can reproduce it, but it's gone in 3.6. Looks like a
> > 3.5 bug, not a bcache bug.
>
> Oh, it was a coincidence then. I hadn't needed to use that for a long
> time.
Yeah. Weird bug.
> > I'm gonna reset the master branch to my current dev branch soon as I've
> > tested it a bit more, but I'd suggest the bcache-3.2 branch - that's the
> > best tested code.
>
> I cannot use a 3.2 kernel. This an HTPC too, besides a NAS and host for
> vm guests. The tuners I have are supported from 3.4 onwards.
>
> Do you mean there might be more problems with the master branch?
Perhaps, but mainly it tracks mainline pretty closely.
> By the way. I have read in the manual how to add new cache devices, but
> can you remove completely the caches in a 'clean way'? Imagine to change
> it to a different disk which for some reason you cannot use at the same time.
Yep. echo 1 > /sys/fs/bcache/<cache set>/unregister
prev parent reply other threads:[~2012-08-21 0:37 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-08-20 7:03 Issues dropping vm fs caches with kernel 3.5.2 + bcache master Javier Marcet
[not found] ` <CAAnFQG9JCMpZ7VSnuNxSGxXa4KrDWjdP4e3UJZj=W2Jbi30dZg-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2012-08-20 20:29 ` Kent Overstreet
[not found] ` <20120820202921.GA21953-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
2012-08-20 21:06 ` Javier Marcet
[not found] ` <CAAnFQG-aK4aR8J05oe=MygbzjkoW4furOSyEjrhw6Zp0vznhEw-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2012-08-21 0:15 ` Kent Overstreet
[not found] ` <20120821001514.GB21953-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
2012-08-21 0:31 ` Javier Marcet
[not found] ` <CAAnFQG_rE9f61Rgo+xH6uNujLwNx_QFaF-OpH_=pYro-+99AFQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2012-08-21 0:37 ` Kent Overstreet [this message]
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=20120821003713.GB19615@google.com \
--to=koverstreet-hpiqsd4aklfqt0dzr+alfa@public.gmane.org \
--cc=jmarcet-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org \
--cc=linux-bcache-u79uwXL29TY76Z2rM5mHXA@public.gmane.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.