Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 03 Mar 2010 10:57:35 +0200
From:      Daniel Braniss <danny@cs.huji.ac.il>
To:        Rick Macklem <rmacklem@uoguelph.ca>
Cc:        stable@freebsd.org, freebsd-fs@freebsd.org, Willem Jan Withagen <wjw@digiware.nl>, =?utf-8?B?RWlyaWsgw5h2ZXJieQ==?= <ltning@anduin.net>, rwatson@freebsd.org, Jeremy Chadwick <freebsd@jdc.parodius.com>
Subject:   Re: mbuf leakage with nfs/udp (was mbuf leakage with nfs/zfs?)
Message-ID:  <E1NmkOe-000PSY-JT@kabab.cs.huji.ac.il>
In-Reply-To: <Pine.GSO.4.63.1003021947470.3879@muncher.cs.uoguelph.ca> 
References:  <20100226174021.8feadad9.gerrit@pmp.uni-hannover.de>  <E1Nl6VA-000557-D9@kabab.cs.huji.ac.il> <20100226224320.8c4259bf.gerrit@pmp.uni-hannover.de> <4B884757.9040001@digiware.nl> <20100227080220.ac6a2e4d.gerrit@pmp.uni-hannover.de> <4B892918.4080701@digiware.nl> <20100227202105.f31cbef7.gerrit@pmp.uni-hannover.de> <20100227193819.GA60576@icarus.home.lan> <BD8AC9F6-DF96-41F9-8E92-48A4E5606DC7@anduin.net> <4B89943C.70704@digiware.nl> <20100227220310.GA65110@icarus.home.lan> <Pine.GSO.4.63.1003011703100.26054@muncher.cs.uoguelph.ca> <E1NmPHy-0009jy-Dj@kabab.cs.huji.ac.il> <Pine.GSO.4.63.1003021947470.3879@muncher.cs.uoguelph.ca>

next in thread | previous in thread | raw e-mail | index | archive | help
> 
> 
> On Tue, 2 Mar 2010, Daniel Braniss wrote:
> 
> > runing with the experimental nfs server all is ok!
> > (at least I can't see any mbuf leakage :-)
> >
> > so now that we can  assume that the problem is in NFS/UDP writes via
> > classic nfsserver, where to look?
> >
> 
> It might also be the krpc reply cache, since the experimental server
> isn't using it (nfsv4 requires a rather twisted reply cache and it was
> easier to just use that one for nfsv2,3 for the experimental server,
> as well).
> 
> >> If it doesn't go away, the problem is more likely in the krpc or the
> >> generic udp code. (When I looked at svc_dg.c, I could only spot one
> >> possible leak and you've already determined that patch doesn't help.
> >> The other big difference when using udp on the FreeBSD8 krpc is the
> >> reply cache code. I seem to recall it's an lru cache with a fixed upper
> >> bound, but it might be broken and leaking.
> >>
> >> If you change the server to set sp_rcache = NULL in the initialization
> >> function in sys/nfsserver/nfs_srvkrpc.c, I think that disables the replay
> >> cache. You wouldn't want to run this way in production, but it would
> >> determine if the leak is in it.
> >>
> >> Change the 3 lines in nfsrv_init() to:
> >> nfsrv_pool->sp_rcache = NULL;
> >> nfsrv_pool->sp_assign = NULL;
> >> nfsrv_pool->sp_done = NULL;
> >>
> >> and I think the krpc replay cache will be disabled.
> >>
> 
> If someone gets a chance to try the above (not in production mode:-),
> it will determine if the problem is in the reply cache or the nfs server's
> write code.
> >> Good luck with it and please report back if you get to try the above.
> >>
> 
> Thanks for trying the experimental server. It is getting narrowed down,
> due to everyone's work on it.
> 
disabling the krpc reply cache does it, no visible damage. Somehow
this reminds me of my old 1970 beetle, parts would fall off but it would
continue working :-)
where to go from here?

	danny





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?E1NmkOe-000PSY-JT>