Date: Fri, 25 Jul 2014 06:32:59 -0400 (EDT) From: Rick Macklem <rmacklem@uoguelph.ca> To: Harald Schmalzbauer <h.schmalzbauer@omnilan.de> Cc: freebsd-stable <freebsd-stable@freebsd.org> Subject: Re: nfsd server cache flooded, try to increase nfsrc_floodlevel Message-ID: <1672084785.3195858.1406284379181.JavaMail.root@uoguelph.ca> In-Reply-To: <53D20A49.5020803@omnilan.de>
next in thread | previous in thread | raw e-mail | index | archive | help
Harald Schmaltzbauer wrote: > Bez=C3=BCglich Rick Macklem's Nachricht vom 25.07.2014 02:14 (localtime): > > Harald Schmalzbauer wrote: > >> Bez=C3=BCglich Rick Macklem's Nachricht vom 08.08.2013 14:20 > >> (localtime): > >>> Lars Eggert wrote: > >>>> Hi, > >>>> > >>>> every few days or so, my -STABLE NFS server (v3 and v4) gets > >>>> wedged > >>>> with a ton of messages about "nfsd server cache flooded, try to > >>>> increase nfsrc_floodlevel" in the log, and nfsstat shows TCPPeak > >>>> at > >>>> 16385. It requires a reboot to unwedge, restarting the server > >>>> does > >>>> not help. > >>>> > >>>> The clients are (mostly) six -CURRENT nfsv4 boxes that netboot > >>>> from > >>>> the server and mount all drives from there. > >>>> > > Have you tried increasing vfs.nfsd.tcphighwater? > > This needs to be increased to increase the flood level above 16384. > > > > Garrett Wollman sets: > > vfs.nfsd.tcphighwater=3D100000 > > vfs.nfsd.tcpcachetimeo=3D300 > > > > or something like that, if I recall correctly. >=20 > Thanks you for your help! >=20 > I read about tuning these sysctls, but I object individually altering > these, because I don't have hundreds of clients torturing a poor > server > or any other not well balanced setup. > I run into this problem with one client, connected via 1GbE (not 10 > or > 40GbE) link, talking to modern server with 10G RAM - and this > environment forces me to reboot the storage server every 2nd day. > IMHO such a setup shouldn't require manual tuning and I consider this > as > a really urgent problem! Well, the default was what worked for the hardware I have to test on: - single core i386 server with 256Mbytes of memory on 100Mbps network Since I have nothing close to 10Gbps networking (100Mbps only), I can't test even a situation like your server/single client, so all I can do is set a default that works for me. I don't think you can expect a "one size fits all" setting when you have servers ranging from what I have (i386 with 256Mbytes of memory) to 64 cores, Gbytes of RAM etc. Eventually, others (like iXsystems maybe) who can test on a range of servers may be able to come up with a way to tune it dynamically based on server size, but that requires a range of hardware to test on. Basically, although NFSv4 is now 10years old, people are just starting to use it, so experience is still pretty limited w.r.t. it. rick > Whatever causes the server to lock up is strongly required to be > fixed > for next release, > otherwise the shipped implementation of NFS is not really suitable > for > production environment and needs a warning message when enabled. > The impact of this failure forces admins to change the operation > system > in order to get a core service back into operation. > The importance is, that I don't suffer from weaker performance or > lags/delays, but my server stops NFS completely and only a reboot > solves > this situation. >=20 > Are there later modifcations or other findings which are known to > obsolete > your noopen.patch (http://people.freebsd.org/~rmacklem/noopen.patch)? >=20 > I'm testing this atm, but having other panics on the same machine > related to vfs locking, so results of the test won't be available too > soon. >=20 > Thank you, >=20 > -Harry >=20 >=20 >=20
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1672084785.3195858.1406284379181.JavaMail.root>