home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!sun-barr!cs.utexas.edu!swrinde!mips!mips!ultra!shj
- From: shj@ultra.com (Steve Jay {Ultra Unix SW Mgr})
- Newsgroups: comp.sys.sun.admin
- Subject: Re: sunrpc mystery needs solving
- Keywords: sunrpc netstat whatthehell
- Message-ID: <1992Jul31.034243.10984@ultra.com>
- Date: 31 Jul 92 03:42:43 GMT
- References: <1992Jul25.012308.346@auto-trol.com> <1992Jul28.074202.24262@fwi.uva.nl>
- Organization: Ultra Network Technologies
- Lines: 40
-
- In <1992Jul28.074202.24262@fwi.uva.nl> casper@fwi.uva.nl (Casper H.S. Dik) writes:
-
- >quejoh@auto-trol.com (Quentin Johnson) writes:
-
- >>I'm trying to understand some weird behavior on a Sun 4/110 running
- >>SunOS 4.1.1.
-
- >>Netstat reports a lively "sunrpc" life for an otherwise sedate little
- >>workstation. The "localhost.sunrpc" entries seem to cycle from none to
- >>hundreds and slowly back down to none. It almost seems as if running
- >>"netstat" causes them to increase!
-
- >This is a side effect of not running NIS and running netstat. Each
- >invocation of netstat tries to connect to the NIS server
- >to resolve port numbers. This involves asking the portmapper about
- >some questions. Since netstat uses tcp (probably by specifying
- >set*ent(stayopen)) you will be left with tha connection in TIME_WAIT
- >state. Don't worry, they'll go away (if you don't look).
-
- Yes, worry. This is actually a nasty bug which Sun should fix. If
- you do repeated netstat's, each one creates several more TIME_WAIT
- sockets for each one that was already there. That is, you increase
- the number of TIME_WAIT sockets by a factor of 4 or 5 every time you
- run netstat. If you do enough netstat's quickly enough, you can fill
- up all the mbuf space in your machine with useless socket structures.
- Never mind that the netstat output becomes pretty useless.
-
- The solution is for the NIS routines to re-use the same socket each
- time they try to re-contact the NIS server, rather than open a new
- one each time. I'm pretty sure the routines already do re-use the
- same socket for talking to a server if it's there. They need to do
- the same thing if the server isn't there.
-
- I guess I should report this to Sun, unless some Sun guy is listening
- on this newsgroup and takes care of this themselves (fat chance).
-
- Steve Jay
- shj@ultra.com ...ames!ultra!shj
- Ultra Network Technologies / 101 Dagget Drive / San Jose, CA 95134 / USA
- (408) 922-0100 x130 "Home of the 1 Gigabit/Second network"
-