home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.sysv386
- Path: sparky!uunet!cs.utexas.edu!torn!skule.ecf!epas!nyama!problem!trinet!uunorth!dlomas!david
- From: david@dlomas.uucp (David Lomas)
- Subject: SCO NFS High Load Averages
- Organization: Lomas Data Management Inc.
- Distribution: na
- Date: Tue, 21 Jul 92 20:10:55 GMT
- Message-ID: <1992Jul21.201055.2579@dlomas.trinet>
- Keywords: FTP+, NFS, Load Average
- Sender: david@dlomas.trinet (David Lomas)
- Lines: 78
-
- Hi all:
- I hope that someone can point me in the right direction with what I'm
- sure is a SCO-related problem.
-
- We're running SCO Unix ver 3.2.2 with TCP/IP & NFS. After a DOS-based
- PC running FTP's TCP+ (ver 1.05) makes any use of NFS (Interdrive),
- the output of the "w" command yields something like;
-
- 8:25am up 2 days, 12:46, 1 users, load average: 4.00, 4.00, 4.00
-
- A partial listing of the output of "ps -ef" gives the following;
-
- root 283 1 0 Jun 17 ? 0:00 slink
- root 296 1 0 Jun 17 ? 0:00 inetd
- root 328 1 0 Jun 17 ? 0:00 portmap
- root 298 1 0 Jun 17 ? 0:01 routed
- root 301 1 0 Jun 17 ? 0:04 rwhod
- root 330 1 0 Jun 17 ? 0:03 pcnfsd
- root 332 1 0 Jun 17 ? 0:01 mountd
- root 335 1 0 Jun 17 ? 5:52 nfsd 4 <<<
- root 337 335 0 Jun 17 ? 5:36 nfsd 4 <<< look at these
- root 338 335 0 Jun 17 ? 6:17 nfsd 4 <<<
- root 339 335 0 Jun 17 ? 5:51 nfsd 4 <<<
- root 340 1 0 Jun 17 ? 0:00 nfsclnt 4
- root 342 1 0 Jun 17 ? 0:00 biod 4
- root 343 1 0 Jun 17 ? 0:00 biod 4
- root 344 1 0 Jun 17 ? 0:00 biod 4
- root 345 1 0 Jun 17 ? 0:00 biod 4
- root 347 1 0 Jun 17 ? 0:00 statd
- root 349 1 0 Jun 17 ? 0:00 lckclnt 4
- root 351 1 1 Jun 17 ? 3:25 lockd <<< and this
- root 353 1 0 Jun 17 ? 0:00 rexd
-
- There's nothing in the SAR reports that show anything extraordinary,
- with the exception of sometimes when the load averages approach 6,
- the CPU %idle tends to 0% (sar -u).
-
- Weird thing is , if no DOS-based PC uses NFS, load averages stay about
- at zero.
-
- If we mount (via NFS) the same part of the SCO system on a Sun system
- via NFS, the load averages (on the SCO box) go up to 2 and stay there,
- the load averages on the Sun box also go up to 1+, but decline to 0
- after "umounting" the SCO drive.
-
- Simply TCP/IP from anybody to anybody, r-login's, and the like have
- no effect.
-
- Our SCO software;
-
- Set Release Notes
- --- ------- -----
- Host Support Module 1.6.6a installed
- (prd = unixos) 3.2.2s partially installed
- NFS Development System 1.1.0b installed
- SCO NFS Runtime System 1.1.1d installed
- SCO UNIX 3.2v2/NFS Maintenance D 1.0.0b installed << lockd fix ??
- SCO TCP/IP Maintenance Supplemen 1.0.0b installed
- SCO LLI Driver disk 1.0.0q installed
- SCO TCP/IP Runtime System 1.1.3f installed
- Quicksilver/UNIX for SCO XENIX 1.3b.2 installed
- Microsoft Word 5.1 5.1.1b installed
- Software Development System 3.2.2a installed
- SCO Professional 2.0.1h removed
- SCO VP/ix 1.2.0n installed
- SCO Computer Graphics Interface 1.1.0d installed
- SCO FoxBASE+ 2.1.1a installed
- NEC Specials 1.0 installed
- (prd = unixos) 3.2.2s installed
- SCO System V Operating System 3.2.2s partially installed
- (prd = unixos) 3.2.2s installed
- (prd = unixos) 3.2.2s partially installed
-
- Sorry for the long posting, if anyone can shed some light ( even heat
- appreciated, sigh...)
-
- --
- David Lomas
-