[Search for users]
[Overall Top Noters]
[List of all Conferences]
[Download this site]
| Title: | VAX DBMS |
| Notice: | THIS NOTESFILE IS NOT A FORMAL SUPPORT CHANNEL |
| Moderator: | SCARY::CHARLAND |
|
| Created: | Thu Feb 20 1986 |
| Last Modified: | Tue Jun 03 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 2642 |
| Total number of notes: | 11044 |
2626.0. "COSI$IO_PUT_OUTPUT +18A" by M5::BLITTIN () Wed May 07 1997 18:17
DBMS 6.1-11
VAX/VMS 6.1
Customer has a batch job that does a DBO/MONITOR REOPEN_LOG weekly.
Job has been running successfully for some time, but this weeks job
*appears* to have caused the monitor process to bugcheck with
***** Exception at 00017CDF : COSI$IO_PUT_OUTPUT + 0000018A
%COSI-F-FILACCERR, error formatting file
-COSI-F-BADPARAM, bad parameter value
There is an identical bugcheck in the RDB_60 notes (4790). I say appears
because the DBMMON.LOG prior to shows the last activity as the REOPEN,
then the new log starts bugchecking.
The bugcheck, once it started, filled the monitor log with repeating
information, about 1.5mg blks worth. The customer tried
DBO/MONITOR STOP, but since they didn't use /ABORT=DELPRC, it
was waiting for existing users to finish. The customer then
STOP/ID on the monitor process, restarted it, and they were back
up.
They want to know what caused this and what can they do to prevent
this from happening again. Their production system was down
for about an hour as a result.
I've got the monitor logs from before and after the bugcheck. If
any of this information would be of value, I can post them here.
So far I haven't been able to reproduce this, nor has the customer.
This is a Workstream environmnet.
Any help would be appreciated...
Thank You...
| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 2626.1 | | HOTRDB::PMEAD | Paul, pmead@us.oracle.com, 719-577-8032 | Wed May 07 1997 19:18 | 1 |
| Do you think you can post the stack dump portion of the bugcheck?
|
| 2626.2 | | HOTRDB::PMEAD | Paul, pmead@us.oracle.com, 719-577-8032 | Thu May 08 1997 17:15 | 7 |
| OK. With the help of my good buddy Norm we spotted the problem. They
have a user on the system that is setting up a process name of
"ByeByeIntegra!". The monitor is not handling the possibility of a "!"
in a process name and things go downhill from there.
As a workaround they can banish the use of "!" in process names. I
will put a fix in the code in the near future.
|
| 2626.3 | Thank You... | M5::BLITTIN | | Thu May 08 1997 18:46 | 2 |
|
Thank you very much Paul & Norm.
|
| 2626.4 | Fixed in 6.1 and later | HOTRDB::PMEAD | Paul, pmead@us.oracle.com, 719-577-8032 | Thu May 08 1997 21:43 | 23 |
| 0.1 Monitor Bugchecks at COSI$IO_PUT_OUTPUT + 18A
The <REFERENCE>(PRODUCT_NAME) database monitor process would
sometimes bugcheck with the following exception:
***** Exception at 0001900D : COSI$IO_PUT_OUTPUT + 0000018A
%COSI-F-FILACCERR, error formatting file
-COSI-F-BADPARAM, bad parameter value
The bugchecks would repeat over and over until the system disk
was full or the monitor exceeded its stack space allocation.
This bugcheck would occur after an RMU/MONITOR REOPEN_LOG com-
mand was issued and only if there was a database user on the
system that had set their process name to include an exclama-
tion point "!". The database monitor had incorrectly interpreted
the exclamation point as a message formatting directive. This
problem would then be exacerbated by a loop condition in the
exception handling for the error that caused the bugcheck to be
written over and over again.
This problem has been corrected in <REFERENCE>(PRODUCT_NAME)
Version <REFERENCE>(VERSION_NUMBER).
|