From: Daniel Barbar Subject: RE: Maximum number of nfsd daemons? Date: Thu, 1 Aug 2002 08:30:40 -0700 Sender: nfs-admin@lists.sourceforge.net Message-ID: Mime-Version: 1.0 Content-Type: text/plain Cc: "'nfs@lists.sourceforge.net'" Return-path: Received: from smtp1.legato.com ([137.69.200.1]) by usw-sf-list1.sourceforge.net with esmtp (Exim 3.31-VA-mm2 #1 (Debian)) id 17aHvg-0005c4-00 for ; Thu, 01 Aug 2002 08:31:56 -0700 To: "'Neil Brown'" Errors-To: nfs-admin@lists.sourceforge.net List-Help: List-Post: List-Subscribe: , List-Id: Discussion of NFS under Linux development, interoperability, and testing. List-Unsubscribe: , List-Archive: Neil, Erich: Thanks for the replies. Here is the content of /proc/net/rpc/nfsd: fiona:~# cat /proc/net/rpc/nfsd rc 19 123835 15603248 fh 10 15227728 0 14 188 io 3752095213 1826934055 th 128 0 34.710 2.230 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 ra 256 8316937 462671 10814 6530 3748 4498 2851 3290 1583 1881 402138 net 15727611 15727611 0 0 rpc 15727102 509 509 0 0 proc2 18 856 35104 6 0 9044 349 329927 0 0 83 0 0 0 0 223 0 1890 569 proc3 22 144698 1357961 50256 519080 2198920 744533 8887018 60150 5539 710 1 0 4486 383 127 0 268983 1005032 88344 6016 4533 2281 Most clients are Solaris (IS&T here is very Solaris centric). I was considering installing the patch to enable NFS/TCP, I just wasn't sure because of its "experimental" state. Have you guys gotten a feeling for its stability? Thanks again, --- Daniel Barbar -----Original Message----- From: Neil Brown [mailto:neilb@cse.unsw.edu.au] Sent: Wednesday, July 31, 2002 9:27 PM To: Daniel Barbar Cc: 'nfs@lists.sourceforge.net' Subject: Re: [NFS] Maximum number of nfsd daemons? On Wednesday July 31, dbarbar@legato.com wrote: > Hi, > > We recently moved one of our NFS servers from a Sun machine running > Solaris 7 to an Intel box running SuSE Linux 8.0 (kernel 2.4.18). We are > experiencing some (rather heavy) performance problems, more so for clients > accesing the service over our WAN link. > I'm starting to go through the excerise of tuning the NFS system, and > started first by trying to increase the number of nfsd process. It seems > that there is a hard coded limit of 128 processes, is that correct? If true, > is it because one shouldn't expect any substantial performance increase by > setting the number of nfsd threads to a number larger than 128? This is correct in 2.4.18. 2.4.18 will have a substantially larger limit. Just change the #define in fs/nfsd/nfssvc.c and recompile for a larger limit. I guess when the number was chosen, computers were smaller. It could be that more than 128 threads will help you, but it would be worth checking /proc/net/rpc/nfsd to see how many threads are really being used. Send me the content of that file and I will interpret bits for you. NeilBrown ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ NFS maillist - NFS@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs