Many thanks to Russ Fish and Mervyn Weis for their fast responses.
We had limited ulimit set to 500M. We took the limitation of and the
database came back up.
Russ suggested to set per-proc-address-space and per-proc-data-size to same
value as
max-per-proc-XXXXX. Or, before starting up the database, in csh - set
limit addressspace unlimited
limit datasize unlimited
Mervyn pointed me to the 4.0D patch site. There is a separate patch
entitled
"threadsafe lib" which, I believe, is related to Oracle and *aio* and
*thread* routines. You might want to download it and the accompanying README
file to determine if installing this patch (manually) will resolve your
issue(s). I would also check to make sure that the modules in the patch are
not already part of PK#5 (I don't believe they are). Nonetheless, install it
(follow the instructions) and then restart the database and observe what
happens.
Martina
----------------------------------------------------------------------
original posting:
Martina Thomas <mthomas_at_siebel.com> on 10/30/99 11:06:05 AM
To: tru64-unix-managers
cc: (bcc: Russ Fish/SEA/IDX1)
Subject: Oracle on DEC Unix 4.0D, patch kit 5
Body:
Hi,
anyone seen this error before and can provide a bit more detail? I cannot
start the database at this point.
SVRMGR> startup
Internal AIO consistency error: Cannot initialize aio library.
Error 35: Operation would block
Aborting...
Thanks,
Martina Thomas
Received on Sat Oct 30 1999 - 19:36:21 NZDT