Hello: I have a 600 5/333 with 1Gb RAM and two 1Gb
swap partitions. When a user starts a job that
allocates about 900Mb of memory, a bit before it has
allocated all of that the system hangs. At this point
there seems to be no recourse but to hit the re-set
button. We have a local office which is our interface
to DEC support and they told me look at the process
limits, so...
root_at_ocelot /sbin/sysconfig -q proc > foo.out
root_at_ocelot more foo.out
max-proc-per-user = 64
max-threads-per-user = 256
per-proc-stack-size = 2097152
max-per-proc-stack-size = 33554432
per-proc-data-size = 134217728
max-per-proc-data-size = 1073741824
max-per-proc-address-space = 1073741824
per-proc-address-space = 1073741824
autonice = 0
open-max-soft = 4096
open-max-hard = 4096
ncallout = 298
ncallout_alloc_size = 8192
round-robin-switch-rate = 0
round_robin_switch_rate = 0
sched-min-idle = 0
sched_min_idle = 0
give-boost = 1
give_boost = 1
And so it appears that there is a per-process limit
on address space of 1Gb and maybe that is the problem.
I am told that I can list values for these variables in
/etc/sysconfigtab (which currently contains nothing)
and if I do a kernel rebuild they will take effect.
I'm not very convinced of this as a solution because
it seems to me that if the problem is these limits
then the particular process should crash with an
"out of core" sort of error. If process limits
are set and reaching those limits hangs the system,
then that is a rather interesting feature of DEC UNIX,
eh? Speaking of which, the machine is running:
Digital UNIX V3.2D-1 (Rev. 41); Tue Feb 13 11:30:27 EST 1996
Digital UNIX V3.2D-1 Worksystem Software (Rev. 41)
Thanks for your consideration, -Korsmeyer
Received on Fri Nov 08 1996 - 16:53:49 NZDT