| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 2534.1 | Time clash?? | COPCLU::ELIN | Elin Christensen @DMO, DTN 857-2406 | Fri Feb 21 1997 13:55 | 72 |
|
I have found the code in SM_AIDA_START_IMAGE.COM that assigns the
process names:
...
$ count = 0
...
$loop:
$ if count .gt. 100 then goto loop_finish
$ count = count + 1
$ on error then goto loop
$ new_count = f$string(count)
$ new_process_name = "''aida$process_name'_''new_count'"
$ set proc/name='new_process_name'
It creates a process name, starting with AIDA$SERVER_1, and adds 1 to
the count till the "set proc/name='new_process_name'" command stops
failing.
As there are no errors logged during AIDA server startup it seems as if
this could be a timing problem.
AIDA$OUTPUT_1996101223100645.LOG_1997022102571815;2
21-FEB-1997 02:57:35.47
AIDA$OUTPUT_1996101223121001.LOG_1997022102571924;2
21-FEB-1997 02:57:35.52
As you can see, the two logfiles - containing the string "Server
starting as process AIDA$SERVER_1" - have been created at almost the same
time.
But some of the other AIDA startup logfiles have been created with even
smaller time differences - without problems!
AIDA$OUTPUT_1996101223080331.LOG_1997022102571683;2
21-FEB-1997 02:57:35.49
AIDA$OUTPUT_1996101223100645.LOG_1997022102571815;2
21-FEB-1997 02:57:35.47
AIDA$OUTPUT_1996101223121001.LOG_1997022102571924;2
21-FEB-1997 02:57:35.52
AIDA$OUTPUT_1996101223125606.LOG_1997022102572743;2
21-FEB-1997 02:57:36.28
AIDA$OUTPUT_1996101223140067.LOG_1997022102572787;2
21-FEB-1997 02:57:39.29
AIDA$OUTPUT_1996101223145143.LOG_1997022102572865;2
21-FEB-1997 02:57:44.57
AIDA$OUTPUT_1996101223155209.LOG_1997022102573099;2
21-FEB-1997 02:57:39.81
AIDA$OUTPUT_1996101223162582.LOG_1997022102573119;2
21-FEB-1997 02:57:44.56
AIDA$OUTPUT_1996101223170488.LOG_1997022102573152;2
21-FEB-1997 02:57:39.80
AIDA$OUTPUT_1996101223201148.LOG_1997022102573180;2
21-FEB-1997 02:57:44.58
As you can see: 02:57:44.56, 02:57:44.57 and 02:57:44.58,
- and 02:57:39.80 and 02:57:39.81.
It is also funny that in between the two AIDA$SERVER_1 logfiles at
02:57:35.47 and 02:57:35.52 another logfile has been created at
02:57:35.49 in which the process name AIDA$SERVER_2 has been assigned!
Comments???
I guess we should stop and restart the AIDA$SERVER_1 servers to be sure
that they run without collisions/problems ????
Elin
|
| 2534.2 | | IOSG::PYE | Graham - ALL-IN-1 Sorcerer's Apprentice | Mon Feb 24 1997 12:25 | 11 |
| Actually I thought that VMS prevented you creating two proceses with
the same name. Are they both running on the same node? Under the same
UIC?
Are both the processes running OK?
Graham
PS There's actually a quicker way to do the 'unique' process name
creation code, using f$context, but it probably still allows a window
for duplication, which I thought VMS reported as an error.
|
| 2534.3 | | ACISS2::LENNIG | Dave (N8JCX), MIG, @CYO | Mon Feb 24 1997 16:32 | 7 |
| It is possible to have two processes with the same name in the same
UIC group. The basic approach is to set your process name while in
one UIC group, then change your UIC to another group.
I don't know if AIDA actually plays with it's UIC on the fly or not...
Dave
|
| 2534.4 | | IOSG::PYE | Graham - ALL-IN-1 Sorcerer's Apprentice | Mon Feb 24 1997 16:55 | 3 |
| Ah yes, I remember about the name only being unique to a group now, but
I didn't think that AIDA *did* change its UIC. The code in the command
procedure doesn't seem to have any obvious holes...
|
| 2534.5 | | COPCLU::ELIN | Elin Christensen @DMO, DTN 857-2406 | Tue Feb 25 1997 11:48 | 18 |
|
Both AIDA$SERVER_1 processes were running on the same node.
They had been started the normal way AIDA servers get started after a
housekeeping procedure.
I did not check the UICs, but I guess the processes ran under username
ALLIN1 as AIDA$SERVER_n processes usually do.
We did not kill the AIDA$SERVER_1 processes, and did not notice any
problems apart from the %OARTL-W-SESSION_ERROR message in the logfile
(see the base note).
After the next night's run of housekeeping there were the usual ten
AIDA servers with unique process names.
- So, no damage done, but we still wonder how this could happen!
Elin
|
| 2534.6 | It happened again! | COPCLU::ELIN | Elin Christensen @DMO, DTN 857-2406 | Wed Apr 16 1997 13:25 | 24 |
|
This week Tele Danmark has two AIDA$SERVER_6 processes on the same node.
There are no warnings or error messages in the AIDA$OUTPUT*.LOG*;* files
for those two AIDA servers.
No users seem to be bothered, but the customer's alarm systems complain
because there is no AIDA$SERVER_10 process.
216343E9 AIDA$SERVER_1 HIB 5 45959 0 00:00:46.78 30331 124
216353EA AIDA$SERVER_2 HIB 5 41083 0 00:00:46.19 29869 116
216343ED AIDA$SERVER_3 HIB 5 45490 0 00:00:49.26 28961 69
21633BF0 AIDA$SERVER_4 HIB 5 48946 0 00:00:52.26 31772 155
2162FBF2 AIDA$SERVER_5 HIB 5 49424 0 00:00:48.06 29576 102
21633BF5 AIDA$SERVER_6 HIB 5 47109 0 00:00:50.00 29393 135
216303F6 AIDA$SERVER_6 HIB 4 43417 0 00:01:02.13 27743 171
2162E3F8 AIDA$SERVER_7 HIB 5 50013 0 00:00:49.14 31175 193
2162F3F9 AIDA$SERVER_8 HIB 5 46426 0 00:00:49.36 30445 72
2162DBFB AIDA$SERVER_9 HIB 5 106887 0 00:01:35.09 43714 238
!!! AIDA$SERVER_10 is missing
I'll open an IPMT.
Elin
|
| 2534.7 | | COPCLU::ELIN | Elin Christensen @DMO, DTN 857-2406 | Wed Apr 16 1997 18:20 | 15 |
|
Just a reply to tell that the OARTL-W-SESSION_ERROR message in the base
note has NOTHING to do with the duplicate process names.
I have found the same message in AIDA$OUTPUT*.LOG* files that are not related
to servers with duplicate names.
The message does not appear in the logfiles related to the two
AIDA$SERVER_6 processes which we have today.
Besides, the OARTL-W-SESSION_ERROR message in the base note is time stamped
7 hours after startup of the server.
Elin
(I still don't know what the OARTL message means.)
|