| Author |
Message
|
| vplehto |
 Posted: Wed Mar 18, 2009 3:31 am Post subject: Messages transfer under message sequence error |
 |
|
Newbie
Joined: 16 Mar 2009
Posts: 6
|
We have noticed that sometimes on atleast WMQ 5.3 under Message sequence number error situation, messages still tend to go through channels.
It seems that atleast small (<2kB) messages go through even continously, although they might take additional 20 seconds to travel.
Has there been any discussions of thiskind anywhere?
Is there any reason why this is happening?
I have tryed too google this but didn't find exact same.
Somewhere was post that in Redbooks there is good info covering sequence error handling / solving. I didn't find this specific Redbook at IBM, so pointer to correct redbook is also apreciated.
VPL |
|
| Back to top |
|
 |
| Vitor |
| Posted: Wed Mar 18, 2009 4:10 am Post subject: Re: Messages transfer under message sequence error |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| vplehto wrote: |
| We have noticed that sometimes on atleast WMQ 5.3 under Message sequence number error situation, messages still tend to go through channels. |
If they do, it's a fault in v5.3 and you should upgrade. If the channel is in doubt because of sequence issues, it should remain in doubt until fixed.
Well you should upgrade even if that's not your problem. WMQv53 is out of support.
| vplehto wrote: |
| It seems that atleast small (<2kB) messages go through even continously, although they might take additional 20 seconds to travel. |
Perhaps it takes 20 seconds for someone to manually resolve the channel?
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| vplehto |
| Posted: Wed Mar 18, 2009 4:34 am Post subject: |
|
|
Newbie
Joined: 16 Mar 2009
Posts: 6
|
This is then fault on 5.3, not exactly sure if this has ever happended on 6.0 at us.
Upgrade is sometimes more painful as handling with unsupported, but these will be upgraded at some time. It's know that these arent supported anymore.
> Perhaps it takes 20 seconds for someone to manually resolve the channel?
This is not case, because we had one application working a quite long time becouse we didn't get alarm on this particular message sequence error. So it worked and the messages slip through, although they were sometimes a little late.
VPL |
|
| Back to top |
|
|
| Vitor |
| Posted: Wed Mar 18, 2009 4:40 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| vplehto wrote: |
| This is then fault on 5.3, not exactly sure if this has ever happended on 6.0 at us. |
I've not seen a sequence number error behave like this on any version of WMQ. If the MCAs get out of sequence, the channel goes to in doubt needs manual intervention. I don't see a mechanism by which a "small" message would get through.
I've seen instances where channels have delivered messages to a target queue in the "wrong" sequence, but that's not an error as WMQ doesn't guarantee delivery sequence will be the same as sending sequence.
| vplehto wrote: |
| Upgrade is sometimes more painful as handling with unsupported, |
No it isn't. And you can get support for upgrade. You can't get any support for an unsupported version (without money).
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| vplehto |
| Posted: Wed Mar 18, 2009 6:19 am Post subject: |
|
|
Newbie
Joined: 16 Mar 2009
Posts: 6
|
The "small" messages slipping through isn't "dreaming" so I can't explain it how or why, but it defenetly is happening. Don't know the exact sizes it passes, might be even a lot bigger.
I might even be able to create this kind of test environment / applications on how to test it. It could then quite easily be tested also on other versions of WMQ similar way.
Upgrade thing would be nice topic/thread on apropriate place to debate/argue on, but I think it's oftopic on this.
The main point in this I think is that this is happening, and It would be also good to find out that this is something that is broken on 5.3 and corrected on later versions. This I will probably have to find out. |
|
| Back to top |
|
|
| exerk |
| Posted: Wed Mar 18, 2009 6:32 am Post subject: |
|
|
Jedi Council
Joined: 02 Nov 2006
Posts: 6339
|
Who has responsibility for the RCVR end of the channels? If not you, how do you know for certain that those responsible are not resetting to the expected value? Vitor's suggestion still stands as valid.
_________________
It's puzzling, I don't think I've ever seen anything quite like this before...and it's hard to soar like an eagle when you're surrounded by turkeys. |
|
| Back to top |
|
|
| kevinf2349 |
| Posted: Wed Mar 18, 2009 6:33 am Post subject: |
|
|
Grand Master
Joined: 28 Feb 2003
Posts: 1311
Location: USA
|
I don't believe the subject of upgrade isn't at all off topic at all. Version 5.3 is out of support. <=== period
However.....are you sure these messages aren't getting to the target queue by some other means? (via another channel, a client connection). Are you sure you don't have automation in place to reset the channel (bad practice, but not unheard of).
Where are you looking at this from, the receiving end? The sending end? or both?
What are you findings to date? Are the sequence numbers still out of whack or do they get resolved?
Have you raised a PMR....(see previous comment about upgrade!) |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Wed Mar 18, 2009 3:55 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
Could those messages that are slipping through be non persistent? 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| vplehto |
| Posted: Thu Mar 19, 2009 4:28 am Post subject: |
|
|
Newbie
Joined: 16 Mar 2009
Posts: 6
|
>exerk
>Who has responsibility for the RCVR end of the channels? If not you,
I'm one of administrating both systems, so no one is reseting and I'm sure on this.
>kevinf2349
>I don't believe the subject of upgrade isn't at all off topic at all. Version >5.3 is out of support. <=== period
On windows upgrading would be fairly easy, but remote being OpenVMS 7.3-2 which doesn't support newer WMQ (it actually has 5.1) and switching to newer OS would change processor architecture that leads to compiling all aplications again. As everything else is still working perfectly there hasn't been forcing reason to upgrade. Even this isn't major.
On a matter of 5.1 5.3, I'm fairly sure that 5.3 <-> 5.3 is doing this also.
Probably will have to test newer and upgrade.
>However.....are you sure these messages aren't getting to the target
>queue by some other means? (via another channel, a client
>connection).
Yes, I'm sure, we usually have two channels for different purposes and these are checked.
>Are you sure you don't have automation in place to reset the channel >(bad practice, but not unheard of).
No, there isn't automation. But now this is caught our attention that something has to be done to this. Probably not to fix this way.
>Where are you looking at this from, the receiving end? The sending
>end? or both?
Both.
>What are you findings to date? Are the sequence numbers still out of
>whack or do they get resolved?
They did get resolved, so only it was odd thing to happend and we didn't find the reason right away.
>fjb_saper
>Could those messages that are slipping through be non persistent?
Could point, they actually are nonpersistent. |
|
| Back to top |
|
|
| kevinf2349 |
| Posted: Thu Mar 19, 2009 5:38 am Post subject: |
|
|
Grand Master
Joined: 28 Feb 2003
Posts: 1311
Location: USA
|
| Does the channel have NPMFAST set to Yes? |
|
| Back to top |
|
|
| vplehto |
| Posted: Thu Mar 19, 2009 6:09 am Post subject: |
|
|
Newbie
Joined: 16 Mar 2009
Posts: 6
|
Yes it has.
Does this and being nonpersistent have something to do with this?
Haven't thought this throughly before.
I'm not familiar with the parameter purpose(will dig out this tomorrow, doesn't have time today anymore). |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Thu Mar 19, 2009 6:19 am Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
NPMSPEED(FAST) tells the message channel agent to send non-persistent messages outside of UoWs. If the msg sequence numbers at both ends don't agree, only your persistent messages be delayed until you resolve the channel.
_________________
I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Thu Mar 19, 2009 9:07 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
| bruce2359 wrote: |
| NPMSPEED(FAST) tells the message channel agent to send non-persistent messages outside of UoWs. If the msg sequence numbers at both ends don't agree, only your persistent messages be delayed until you resolve the channel. |
Experience shows that the time needed for non persistent message on NMSPEED fast to travel the channel when there is an out of sequence event is far superior than when the sequence numbers match, to the point that most of them will be outside the SLA. The channel will also briefly go into retrying. This is why it is always best to solve those pesky little problems ASAP.
_________________
MQ & Broker admin |
|
| Back to top |
|
|
|
|
