home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!olivea!spool.mu.edu!sdd.hp.com!zaphod.mps.ohio-state.edu!swrinde!emory!gatech!rutgers!micro-heart-of-gold.mit.edu!bu.edu!bu-pub.bu.edu!jdh
- From: jdh@bu-pub.bu.edu (Jason Heirtzler)
- Newsgroups: comp.sys.sgi
- Subject: Re: select: resource temporarily not available
- Message-ID: <106192@bu.edu>
- Date: 5 Jan 93 16:51:33 GMT
- References: <106115@bu.edu> <1993Jan4.232438.16078@odin.corp.sgi.com>
- Sender: news@bu.edu
- Reply-To: jdh@bu-pub.bu.edu (Jason Heirtzler)
- Organization: Boston University, Information Technology
- Lines: 50
-
-
- |>In <106115@bu.edu> jdh@bu-pub.bu.edu (Jason Heirtzler) writes:
- |>>We're having some sort of streams related problem
- |>>with 4.0.5 on our 4D-240, which has been going on since
- |>>at least 3.3 (I probably reported it to the hotline
- |>>even before that release, but that's not what I want to
- |>>get into..)
- |>
- |>>NPROC has been increased to 256 to no avail. I also
- |>>mentioned that I'd try increasing NSTREVENT but that
- |>>I thought that wouldn't help, as something is probably
- |>>eating the resource and not freeing it. Well, I tried
- |>>increasing it from 30 -> 60 -> 120 and it makes no
- |>>difference, the machine still runs out of "events."
- |>
- |>Perhaps you can try to increase your streams buffer parameters.
- |>You have a pool of NBLK buffers (in your /usr/sysgen/master.d/kernel
- |>file, ranging from 4096 to 4 byte data blocks), which you can try
- |>increasing. These data blocks are used for your streams operations.
- |>If you run 'netstat -m', you'll see how many of each pool you are
- |>using, and which need to be increased. And as always when building
- |>a new kernel, make sure to save your old kernel as a backup to boot
- |>from.
-
- Are the NBLK* buffers reported by "netstat -m" different than
- "events"? There are no failures reported under NBLK in netstat.
-
- Is there any kernel debugging information (dbx -k?) to determine
- what's using up all the events?
-
- I was hoping someone from the SGI kernel group would speak up and
- offer a few sage words? Maybe the omnipresent D.O. could comment?
-
- |>If you have more problems, give the hotline folks a call, so that
- |>we can look at your problem in more detail.
-
- What I was aluding to at the start of my article (ahem!) is that
- the hotline hasn't been able to solve it in the past, so I'm sort
- of reluctant to get into that whole mess again. Don't take this
- an attack on the hotline (as there are some good people there.)
-
- This bug has gotten bad enough that users are starting to flip the
- power switch out of frustration, blowing everyone out of the water.
-
- ----------------------------------------------------------------
- Jason Heirtzler 1-617-353-2780 jdh@bu-pub.bu.edu
- Sci Comp & Vis Information Technology Boston University
-
- "Congratulations. You aren't running Eunice."
- -- Perl "Configure" script
-