🚀 go-pugleaf

In article <1993Jan27.144336.14130@cs.hw.ac.uk>, raz@cs.hw.ac.uk (Ross A Macintyre) writes:
>
>Recently errors like these have been reported on our 5000/200.
>These errors are reported by users and don't seem to appear in the error logger.
>This was the output after "ls -l":
>
>*bug.chk.sgi: File table overflow
>*yp_pass.chk: File table overflow
>*extra_src: File table overflow
>*total 546
>*getsvc: stat of /etc/svc.conf failed
> [ etc, etc, etc, ]

The command of interest is pstat(8) with the option -T.  This prints
the sizes of some of the kernel tables.  I'll include sample:

% pstat -T
 171/ 609       files
 138/ 444       gnodes
  48/ 288       processes
  34/  42/  76  active/reclaimable/total texts
 253/1843       00k swap

You seem to be running of entires in both the "files" and "gnodes" table
and probably need to make both larger.  The simplest way is to simply
raise "maxusers" in the configuration, rebuild and reboot.  The sizes
of these tables are calculated based on the value of maxusers.  This
will also increase the size of some of the other tables.  Now most of
the entires are pretty small and it would a large increase of maxusers
to use signifncantly more memory for the tables.  Still, if this is a
concern, the sizes are set in param.c and you could modify the algorithm
that calculates the sizes.

>
>Thanks in advance,
>
>Ross
>
>PS The only real change to the system in the last few days is that I added 
>automount maps to Yellow Pages.
>
--
Alan Rollow				alan@nabeth.cxo.dec.com