Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 26 Feb 2010 16:37:48 +0100
From:      Willem Jan Withagen <wjw@digiware.nl>
To:        Daniel Braniss <danny@cs.huji.ac.il>
Cc:        stable@freebsd.org, =?ISO-8859-1?Q?Gerrit_K=FChn?= <gerrit@pmp.uni-hannover.de>, Jack Vogel <jfvogel@gmail.com>, Jeremy Chadwick <freebsd@jdc.parodius.com>
Subject:   Re: em0 freezes on ZFS server
Message-ID:  <4B87EACC.80105@digiware.nl>
In-Reply-To: <E1Nl1mb-0002Mx-M9@kabab.cs.huji.ac.il>
References:  <4B86F384.3010308@digiware.nl> <2a41acea1002251459v40e8c6ddxd0437decbada4594@mail.gmail.com> <4B8795B1.4020006@digiware.nl> <20100226120339.GB17798@icarus.home.lan> <20100226133138.d47dd080.gerrit@pmp.uni-hannover.de> <20100226134429.041ea6f2.gerrit@pmp.uni-hannover.de> <E1Nkzry-0000AC-0S@kabab.cs.huji.ac.il> <20100226141754.86ae5a3f.gerrit@pmp.uni-hannover.de> <E1Nl1mb-0002Mx-M9@kabab.cs.huji.ac.il>

next in thread | previous in thread | raw e-mail | index | archive | help
On 26-2-2010 16:07, Daniel Braniss wrote:
>> On Fri, 26 Feb 2010 15:04:37 +0200 Daniel Braniss<danny@cs.huji.ac.il>
>> wrote about Re: em0 freezes on ZFS server :
>>
>> DB>  >  At least in my case I found out what is eating the buffers: nfsd
>> DB>  >  does! The buffers stop increasing as soon as I stop nfsd. However,
>> DB>  >  they start increasing as soon as I start nfsd again.
>> DB>  >  Are there any ideas how to fix this? Downgrading back to 7-stable is
>> DB>  >  not really an easy task as far as I know, and I need the server to
>> DB>  >  run without having to reboot it once for twice a day...
>>
>> DB>  I want to add some spices to this stew: :-)
>>
>> You're welcome. :-)
>>
>> DB>  Some few day later it hung, and it's now hanging every few days.
>> DB>  Most of the hangs are because there is no network, but the NIC is bce
>> DB>  not em! I doubled kern.ipc.nmbclusters and lets see what happens ...
>>
>> Do you have nfsd running and serving clients? If so, we should maybe
>> change the topic to something like "possible nfs mbuf leakage"...
>>
> it's only purpose in life is a nfs server.
> but I wouldn't exclude zfs from the equation yet.
> I have othere nfs servers, not doing zfs and I don't see this.
>
>> DB>  23066/6634/29700 mbufs in use (current/cache/total)
>>
>> My server is at 22k now, and the buffer number is still increasing every
>> few seconds...
>> Can you monitor your mbuf usage and report if it grows?
>>
> I am, and in the last 2hs. it grew by about 300, it does oscilate, i.e. it
> grows some, then
> it goes down, but it seems that the low always increases.
>
> when I have enough data i'll plot it.

Same here, from the odd samples I make.

If I don't run my compiles, but just let the server sit.
It just oscilates +- 500 mbufs.

Once I start compiling, mbufs starts growing. I'm now at 22K

Main network (NFS) traffic is than
	postfix writing Maildir
	dovecot - Maildir work

Next to that it is also Samba server, but that doesn't seem to matter that 
much.....

--WjW



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4B87EACC.80105>