On 19/06/2022 10:18 am, kekronbekron wrote:
Would using DISP=SHR,DSN=xx,FREE=CLOSE in the 15-min writer STC work?
Or just YOLO it like the object storage people, i.e., 1 file for 15 minutes in
USS.
Or, if you're willing to explore a little, check out NATS.
It's a light-weird 'MQ' of the modern age. NATS CLI should save you from most
of the setup trouble.
Don't know if it'll work in USS; but since it's Go, you may be able to get it
going with Open Enterprise SDK for Go.
Yeah, good luck with that. My experience of trying to build Prometheus
on z/OS was underwhelming. The syscall API is very different to Linux so
you end up
with a rather large porting effort no dissimilar to C/C++. I abandoned
it. Stick to the JVM, notwithstanding JNA (which I believe is being
ported by IBM)
For a lightweight MQ (or Kafka) you can use Redis, which is used as a
message broker, has pub/sub fanouts and streams similar to Kafka. I
ported it a a decade ago and it
was made public a couple of years back by one of our Zowe devs when it
was going to be used for the caching layer in Zowe. I patched it for
both EBCDIC and enhanced
ASCII and added some z/OS specific features like an MVS STOP command
handler.
[1] https://github.com/lchudinov/redis
- KB
------- Original Message -------
On Sunday, June 19th, 2022 at 2:35 AM, Paul Gilmartin
<0000042bfe9c879d-dmarc-requ...@listserv.ua.edu> wrote:
On Sat, 18 Jun 2022 12:18:12 -0700, Charles Mills wrote:
That's a thought. I had not thought of that. I am REAL familiar with SMF
including user SMF records. APF authorization is not a problem -- I am the
master of this universe.
I suspect that mixing MVS and UNIX technologies might embody the worst
characteristics of both. Try doing repeated fwrite() of small records to a
data set with large BLKSIZE while doing "tail -f" from another job.
The KSDS method sounds promising. I'm entirely VSAM-naive.
Thinking further on the idea of a Rexx logger, I don't know that there's a
good way to handle the SIGHUP. And there's still the hazarrd that the
SIGHUP arrives right during the write(). Serialization required. That's
a job for semaphores. I see no Rexx SYSCALL semaphore support.
But the logger itself could rotate the log files itself, daily. And thee
reporter
could run a few minutes later by crontab; verify that the log has ben
renamed (error handling?) and generate the report. Or the logger itself
could generate the report once daily.
--
gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
