My try at Perl 5.8 RC2 finished early this morning (Sat.). I didn't have
get the spawn error.
I'm not setting any LE runtime opts explicitly, I think we just use the
installation defaults.
Here are a couple of relevant environment vars that I do have set:
_BPX_SHAREAS="YES"
_BPX_SPAWN_SCRIPT="YES"
Here are our system's OMVS settings.
OS/390 UNIX CURRENT CONFIGURATION SETTINGS:
MAXPROCSYS = 4096 MAXPROCUSER = 7500
MAXFILEPROC = 1024 MAXFILESIZE = NOLIMIT
MAXCPUTIME = 2147483647 MAXUIDS = 20
MAXPTYS = 256
MAXMMAPAREA = 16777216 MAXASSIZE = 2147483647
MAXTHREADS = 200 MAXTHREADTASKS = 200
MAXCORESIZE = 4194304 MAXSHAREPAGES = 32768000
IPCMSGQBYTES = 1048576 IPCMSGQMNUM = 20000
IPCMSGNIDS = 4096 IPCSEMNIDS = 4096
IPCSEMNOPS = 32767 IPCSEMNSEMS = 32767
IPCSHMMPAGES = 25600 IPCSHMNIDS = 4096
IPCSHMNSEGS = 1000 IPCSHMSPAGES = 786432
SUPERUSER = BPXROOT FORKCOPY = COW
STEPLIBLIST =
USERIDALIASTABLE=
PRIORITYPG VALUES: NONE
PRIORITYGOAL VALUES: NONE
MAXQUEUEDSIGS = 1000 SHRLIBRGNSIZE = 67108864
SHRLIBMAXPAGES = 4096 VERSION = /
SYSCALL COUNTS = NO TTYGROUP = TTY
SYSPLEX = NO BRLM SERVER = N/A
LIMMSG = NONE
"Van Duser, Brad"
<brad.vanduser@ed To: [EMAIL PROTECTED],
[EMAIL PROTECTED]
s.com> cc:
Subject: RE: second release candidate
for perl 5.8.0 now available
06/27/2002 10:32
PM
Peter, thanks for the clarification on the compile errors and warnings
about
pod2man. I will ignore those for now.
Unfortunately, what I thought was an MVS system problem encountered during
my "make test" appears to be a problem at _test_tty. My make test is
basically dead at "./perl TEST" as shown in this list (pid #724):
PID PPID COMMAND
33554993 83886639 make test
16777960 33554993 /bin/sh -c PERL=./perl make _test
16777958 16777960 make _test
16777957 16777958 /bin/sh -c if (true </dev/tty) >/dev/null 2>&1; then
make TEST_ARGS= TESTFILE
33555171 16777957 make TEST_ARGS= TESTFILE=TEST _test_tty
16777964 33555171 /bin/sh -c cd t && /bin/env ./perl TEST </dev/tty
724 16777964 ./perl TEST
What happens is that when pid 724 starts we get this in syslog:
BPXP005I A FORK OR SPAWN ERROR WAS ENCOUNTERED. RETURN CODE 00000070
REASON CODE 0B15001E
and then pid 724 keep trying to re-create the child process about every 5
seconds and fails with this:
BPXP012I FORK SYSCALL TERMINATED DURING CHILD PROCESSING WITH 650
RETURN CODE 00000070, REASON CODE 00000004.
THE CHILD PROCESS IS 67109626 IN ASID 00A4.
THE PARENT PROCESS IS 724, UID 0, IN ASID 0092.
until I kill it. The return code 70 indicates a temporary resource
shortage. The reason 0004 in undocumented and I opened an IBM ticket on
that but the initial 001E reason code is documented. It states that the
fork failed due to not having enough "work space" in the parent process and
it lists some examples. The fixes are to reduce the number of tasks and/or
the number of load modules (programs).
I have included all of this information to IBM in my ticket, however, they
may come back and tell me that the problem lies within Perl and that
additional work on this issue is billable (which my company won't cover).
If they will at least tell me which resource is short we may be to
determine
if a setting change can correct this.
We were running a little low on the LE HEAP settings so I increased them
all
the way up to what's specified in README.os390. I have also displayed the
OS/390 UNIX resources using 'D OMVS,L' and we are way under the maximum
limits.
I am interested if anyone else has built Perl 5.8 RC2 on an OS/390 system,
specifically OS/390 2.10.
Thanks,
Brad Van Duser
EDS - BellSouth Account
MS 6J72 BSC
675 West Peachtree Street
Atlanta, GA 30308
* phone: +01-404-529-2923
* mailto:[EMAIL PROTECTED]
* pager: +01-800-862-0399
* pager mailto:[EMAIL PROTECTED]
