Re: Dropped users on AIX 5.1
Steve Weet had processes dying on AIX. John Wheeler reported similar disconnections on IDS 7.31UC5/AIX after a hardware upgrade and fixed them by using TCP/IP instead of shared memory. Steve removed onmode -F from cron and adjusted shared memory parameters, and had switched connections to onipcshm. Fred Prose agreed onmode -F is risky before 9.x. Ronald Cole asked about AIX killing processes under memory pressure. Not confirmed fixed.
Auto-generated by Claude from the posts below — may be imperfect; read the full thread.
steve@weet.demon.co.uk (Steve Weet) wrote in message news:<b01c86f1.0407150119.aaf63bc@posting.google.com>... > We are experiencing an issue where 4gl programs (4js) are exiting with > no error logging of any kind. The processes just die. This is > happening to all user processes that have a database connection, > although our menu program (Which has no database connection is not > dying) > When you say the process dies, do you mean it hangs but does nothing or does the process actually end? We were experiencing disconnections from the database with processes hanging and not logging errors. This was after a hardware upgrade from IBM P660 to P690. Seems for some reason connecting to Informix via shared memory wasn't always happy on faster processors. We changed the connection to be via tcpip, works a treat now. I4GL application, IDS 7.31UC5, AIX 5.2
John Wheeler wrote:
> steve@weet.demon.co.uk (Steve Weet) wrote in message news:<b01c86f1.0407150119.aaf63bc@posting.google.com>...
>
>>We are experiencing an issue where 4gl programs (4js) are exiting with
>>no error logging of any kind. The processes just die. This is
>>happening to all user processes that have a database connection,
>>although our menu program (Which has no database connection is not
>>dying)
>>
>
>
> When you say the process dies, do you mean it hangs but does nothing
> or does the process actually end?
>
> We were experiencing disconnections from the database with processes
> hanging and not logging errors. This was after a hardware upgrade from
> IBM P660 to P690. Seems for some reason connecting to Informix via
> shared memory wasn't always happy on faster processors. We changed the
> connection to be via tcpip, works a treat now.
>
> I4GL application, IDS 7.31UC5, AIX 5.2
Thanks for that John. The process actually dies. I have come to the conclusion
that it is the onmode -F that is causing the problem and I have removed that
from cron and adjusted the Shared memory paramaters so hopefully it won;t be needed.
Interesting comment about the shared memory vs networks as this is a new and
quick machine but there seems to be some network latency so we have switched
from onsoctcp to onipcshm. I will bear tihs is mind if we are still getting
issues today.
Steve
I think you are wise in removing the "onmode -F". I always ran into
problems using that command in pre 9.x on AIX. I would often loose
the instance.
Adjusting memory values to provide what Informix needs from the
beginning is always safer.
Steve Weet <sweet@dciltd.com> writes:
> John Wheeler wrote:
>> steve@weet.demon.co.uk (Steve Weet) wrote in message news:<b01c86f1.0407150119.aaf63bc@posting.google.com>...
>>
>>>We are experiencing an issue where 4gl programs (4js) are exiting with
>>>no error logging of any kind. The processes just die. This is
>>>happening to all user processes that have a database connection,
>>>although our menu program (Which has no database connection is not
>>>dying)
>>>
>> When you say the process dies, do you mean it hangs but does nothing
>> or does the process actually end?
>> We were experiencing disconnections from the database with processes
>> hanging and not logging errors. This was after a hardware upgrade from
>> IBM P660 to P690. Seems for some reason connecting to Informix via
>> shared memory wasn't always happy on faster processors. We changed the
>> connection to be via tcpip, works a treat now.
>> I4GL application, IDS 7.31UC5, AIX 5.2
>
> Thanks for that John. The process actually dies. I have come to the
> conclusion that it is the onmode -F that is causing the problem and I
> have removed that from cron and adjusted the Shared memory paramaters
> so hopefully it won;t be needed.
>
> Interesting comment about the shared memory vs networks as this is a
> new and quick machine but there seems to be some network latency so we
> have switched from onsoctcp to onipcshm. I will bear tihs is mind if
> we are still getting issues today.
Does AIX 5.1 still indiscriminately kill processes when under heavy
memory pressure as it predecessors did?
--
Forte International, P.O. Box 1412, Ridgecrest, CA 93556-1412
Ronald Cole <ronald@forte-intl.com> Phone: (760) 499-9142
President, CEO Fax: (760) 499-9152
My GPG fingerprint: C3AF 4BE9 BEA6 F1C2 B084 4A88 8851 E6C8 69E3 B00B