| Author |
Message
|
| gbarber |
 Posted: Thu Jan 16, 2020 1:37 pm Post subject: V6.0 JMS Client subscriber receiving corrupt message |
 |
|
Novice
Joined: 16 Jan 2020
Posts: 10
|
New to the site and think we have a doozy on our hands.
We have a V9.1 QM With PSMODE(ENABLED)
We have a publisher publishing an XML message containing a large (>260KB) base64 encoded string.
After creating a new subscription a v5.3 or v6.0 JMS client receives the message content as it was published. Subsequent receipts of the same message (whether or not other messages are received in between) result in the subscriber receiving a corrupt copy of the message.
The corruption is quite alarming because the received message is truncated form of the published message and the remainder of character positions are filled in with the content from other messages that we know the broker is handling. So, the message received is the same length as the original message, but not the correct content.
NOTE1: Our V9.1 JMS client subscribers receive the message without corruption.
NOTE2: The V5.3 and V6.0 JMS subscribers receive the corrupted form of the message regardless of whether the publisher is V5.3, V6.0 or V9.1
NOTE3: We do not see this corruption when other similarly structured messages that contain even longer, but different base64 encoded strings, are published.
NOTE4: We are able to reproduce this problem in our test environment by subscribing and publishing with an in-house JMS based pub/sub utility application. For this investigation we used both the V5.3 and V6.0 JMS jars to recreate the problem. We have not been able to identify what is special about this particular message.
NOTE5: We ran the subscriber with JMS trace logging turned on, so we have those logs to look at. The message appears corrupted in the trace logs also.
Appreciate any help the group can offer on this one.
Thanks |
|
| Back to top |
|
 |
| Vitor |
| Posted: Thu Jan 16, 2020 1:43 pm Post subject: Re: V6.0 JMS Client subscriber receiving corrupt message |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| gbarber wrote: |
| Appreciate any help the group can offer on this one. |
Don't use V5.3 or V6 anything. They're hideously out of support and you might see some weird effects mixing that with V9.1......
......oh wait......
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Thu Jan 16, 2020 5:18 pm Post subject: Re: V6.0 JMS Client subscriber receiving corrupt message |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
| gbarber wrote: |
| ...a corrupt copy of the message |
In what way(s) is the message corrupt?
_________________
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 |
|
|
| gbarber |
| Posted: Fri Jan 17, 2020 7:06 am Post subject: |
|
|
Novice
Joined: 16 Jan 2020
Posts: 10
|
We're unable to upgrade our JMS clients at this time so we need to deal with the problem at hand. Based on IBMs compatibility documentation the V6 client is supported with V9.1 QM so this should definitely not be happening.
Regarding the type of corruption we're seeing...
The message is truncated to a consistent point and then the contents of another message is appended in place of the remaining character locations. The "other" message who's content is appended is not consistent and is just some message that we know the QM handled recently. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Fri Jan 17, 2020 7:19 am Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
| gbarber wrote: |
We're unable to upgrade our JMS clients at this time so we need to deal with the problem at hand. Based on IBMs compatibility documentation the V6 client is supported with V9.1 QM so this should definitely not be happening.
Regarding the type of corruption we're seeing...
The message is truncated to a consistent point and then the contents of another message is appended in place of the remaining character locations. The "other" message who's content is appended is not consistent and is just some message that we know the QM handled recently. |
Have you verified that the published message is correct?
Assumptions
- The message gets corrupted before the put. MQ is fine. You're just publishing garbage
- The message before the put is fine. An admin subscription and RFHUtil show the corrupted message
- The message in the admin subscription is fine. It is only when retrieving the dynamic subscription that you see the corruption.
Determine which of the above cases matches your scenario. After that you can look into more.
Just be advised that the JMS support pack for MQ 5.3 is way obsolete and may not render correctly anything any more. I would suggest that you set the properties to RFH2 and not Compat for anything below MQ 7.5.
Hope this helps 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Fri Jan 17, 2020 7:28 am Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
| gbarber wrote: |
Regarding the type of corruption we're seeing...
The message is truncated to a consistent point |
Consistent how? On a nKB boundary?
| gbarber wrote: |
| and then the contents of another message is appended in place of the remaining character locations. The "other" message who's content is appended is not consistent and is just some message that we know the QM handled recently. |
FJ's suggestion is spot-on: Where does the corruption occur?
_________________
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 |
|
|
| Vitor |
| Posted: Fri Jan 17, 2020 8:26 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| gbarber wrote: |
| We're unable to upgrade our JMS clients at this time so we need to deal with the problem at hand. |
If not now, when? You've been sitting on those for a while
| gbarber wrote: |
| Based on IBMs compatibility documentation the V6 client is supported with V9.1 QM so this should definitely not be happening. |
I'm assuming you mean this. Given that this is a supported configuration, a PMR is an option.
| gbarber wrote: |
Regarding the type of corruption we're seeing...
The message is truncated to a consistent point and then the contents of another message is appended in place of the remaining character locations. The "other" message who's content is appended is not consistent and is just some message that we know the QM handled recently. |
That sounds like poor memory management & a prime candidate for a PMR given the supported nature of the configuration. You may need an APAR / fix pack as detailed in that link.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| gbarber |
| Posted: Tue Mar 17, 2020 1:14 pm Post subject: CLOSED: IBM acknowledged this as being a bug |
|
|
Novice
Joined: 16 Jan 2020
Posts: 10
|
IBM support indicated that this would be resolved in 9.1.0.6
They also indicated that the bug only manifests if the qm.ini has
MQIBindType=FASTPATH |
|
| Back to top |
|
|
|
|
