| Author |
Message
|
| parnaud |
 Posted: Fri May 10, 2002 1:07 pm Post subject: |
 |
|
Newbie
Joined: 09 May 2002
Posts: 3
Location: SAIC
|
HI, all... new to the group so, if not in the right spot I apologise...
Here is the problem.
I'm using the trace node as a basic msg flow monitoring.
The trace node simply states> time is: ${CURRENT_TIMESTAMP}
The message flow is also basic:
Input node -> trace Node -> exec node (does not much) -> output node
The trace node writes to a file.
The file time stamp info is as follows (writing the relevant info here)
10:17:58.45645 (this is the time info of relevance here 10am 17 minutes etc)
10:17:59.56564
10:17:02.02321
10:17:03.07476
Eventually the minutes tick over to 18.
The question is: what's happening here? I'm running on AIX 4.3.2
I have not seen any support pac mentioning this kind of error (if it is one)
The message flow has around 100 msgs pumped into the input queue (no throttling)
I tried 50 messages -> made no difference to behaviour.
Any help would be appreciated on that one... Using the extract function also made no difference.
Thanks....
_________________
Kind Regards
Philippe |
|
| Back to top |
|
 |
| kirani |
| Posted: Fri May 10, 2002 8:28 pm Post subject: |
|
|
Jedi Knight
Joined: 05 Sep 2001
Posts: 3779
Location: Torrance, CA, USA
|
I am not able to get your problem here. Are you saying that "time is:" string is not getting printed in the file?
What version of MQSI/WMQI are you using and what CSD level?
_________________
Kiran
IBM Cert. Solution Designer & System Administrator - WBIMB V5
IBM Cert. Solutions Expert - WMQI
IBM Cert. Specialist - WMQI, MQSeries
IBM Cert. Developer - MQSeries
|
|
| Back to top |
|
|
| parnaud |
| Posted: Mon May 13, 2002 6:56 am Post subject: Sorry, more explanations |
|
|
Newbie
Joined: 09 May 2002
Posts: 3
Location: SAIC
|
Thanks, here is what I know about the versions:
MQSI: 2.0.2.0
MQSeries: 5.2.0.0
CSD level is unknown, I need to get in touch with the Environment guys and see what they installed. The next question is what should it be? (as in: which one fixes bugs that cause headaches, rather than more functionality enhancements shall we say).
The trace node is set as follows:
Write to: file
Destination: /tmp/inbound.log
Details:
Time is: ${CURRENT_TIMESTAMP}
I would expect it to be interpreted as STRING("Time is: ") + STRING( converted timestamp in ascii format).
Details is simply the current time, I took this from an example in the documentation (ok, sometime documentation is known for not being entirely accurate in citing examples).
The log file '/tmp/inbound.log' is created and shows the following text:
...
Time is: 10:17:58.456457
Time is: 10:17:59.565684
Time is: 10:17:02.232613
Time is: 10:17:03.747638
...
I tried to see what CSD were released and bugs/features they fixed or provided. I'm running on AIX. Since this seems a pretty obvious bug, I am of the opinion that I'm doing something pretty stupid, except it's like the missing ';' case... I just cannot see it/work it out!
I've tried using EXTRACT(minute,...) and EXTRACT(second,...) but the fault persisted.
The message flow executes as 'single thread' so I do not expect 'race condition' between concurrently executing flows.
I hope this helps a bit...
Kind regards
Philippe |
|
| Back to top |
|
|
| kolban |
| Posted: Mon May 13, 2002 7:11 am Post subject: |
|
|
Grand Master
Joined: 22 May 2001
Posts: 1072
Location: Fort Worth, TX, USA
|
Oooh ... I love this ... love it 
By any chance do you either have multiple instancies of the message flow deployed to multiple execution groups or have you tagged the message flow to have multiple instancies? If so, then that may explain the results. Check the open input count on the queue from which the messages are bveing retrieved. It should be one. If greater than that, we may have the situation described above. |
|
| Back to top |
|
|
| parnaud |
| Posted: Tue May 14, 2002 1:05 pm Post subject: TimeStamp - still no go |
|
|
Newbie
Joined: 09 May 2002
Posts: 3
Location: SAIC
|
Sorry it took a while to get back but our environment went down.
We are operating in 'single thread' mode. I double checked that and ensured that the input msg is only retrieved after the output queue has been updated with the outgoing message.
Still, no luck to try and figure this one. I thought I tried the ExecuteTime node and see what it says about the elapsed time and compare it to the trace nodes time information (ok these should be quite close). I double checked the config info for syslog and also ensured that the syslog file (for user.info) was properly setup... Would you believe I get messages in that file except the one indicating the elapsed time from the ExecuteTime node... I know information message go there and, according to the release note for the node, so should the timing information. All I can say is now I am completely and absolutely confused... Any one could oblige me and tell me the steps I should follow for the proper trace message display/write for the ExecuteTime node? If that does not work, I'll have to assume the testing environment is gone awry and may need to be rechecked (if not reinstalled / reconfigured from scratch).
Thanks again....
Kind regards
Philippe (The Confused One) |
|
| Back to top |
|
|
| joebrew |
| Posted: Thu May 16, 2002 6:34 am Post subject: I'm having a similar problem |
|
|
Newbie
Joined: 16 May 2002
Posts: 5
Location: Harrisburg, PA
|
I'm experiencing similar problems with the CURRENT_TIMESTAMP on AIX.
In my first message flow, I insert a record into an Oracle database and set a create_timestamp field = CURRENT_TIMESTAMP.
My next message flow updates the same record in an update_timestamp field and sets the field = CURRENT_TIMESTAMP.
I am seeing situations where the update_timestamp field is earlier than the create_timestamp field. It always seems to happen around the turn of the minute.
For example, create_timestamp = 05/16/2002 10:03:59
and the update_timestamp = 05/16/2002 10:03:00
In my scenario, the minutes seem to increment before the seconds. I.E. the create timestamp should be 10:02:59 instead of 10:03:59.
Here is an excerpt of my ESQL:
FIRST FLOW:
DECLARE curTS TIMESTAMP;
SET curTS = CURRENT_TIMESTAMP;
INSERT INTO Database.STEP(STEP_SEQ_ID, CRT_TS)
VALUES
CAST( CAST( Root.XML.Message.AddStep.Step_Seq_Id AS FLOAT ) AS INT ),
SUBSTRING( CAST( curTS AS CHAR ) FROM 12 FOR 19 ));
The Oracle database timestamp fields are defined as date.
SECOND FLOW:
DECLARE curTS TIMESTAMP;
SET curTS = CURRENT_TIMESTAMP;
UPDATE Database.STEP AS T
SET UPDT_TS =SUBSTRING( CAST( curTS AS CHAR ) FROM 12 FOR 19 )
WHERE T.STEP_SEQ_ID = CAST Root.XML.EndStep.Step_Seq_Id AS INT)
Any help would be appreciated.
Thanks - Joe |
|
| Back to top |
|
|
| joebrew |
| Posted: Mon May 20, 2002 11:18 am Post subject: IBM Response |
|
|
Newbie
Joined: 16 May 2002
Posts: 5
Location: Harrisburg, PA
|
Here is the response from IBM:
After detail search we found a similar problem with one of our other
Customer.
This problem is fixed in csd02 (problem IY19242 ).
So we recommend you to apply csd02,which should the solve the issue.
Pl let us know the status after applying csd02. |
|
| Back to top |
|
|
| joebrew |
| Posted: Thu May 23, 2002 10:59 am Post subject: IT WORKED! |
|
|
Newbie
Joined: 16 May 2002
Posts: 5
Location: Harrisburg, PA
|
| After applying CSD02, we are no longer experiencing the problem. |
|
| Back to top |
|
|
|
|
