| Author |
Message
|
| RaviKrG |
 Posted: Fri Jul 27, 2012 9:57 am Post subject: Queue Manager Goes for a Hang |
 |
|
Master
Joined: 07 Sep 2008
Posts: 240
|
Hi, we have a qmgr running on Solaris since very long time but now we are facing a problem which shows that queue manager is getting hung frequently.
MQ : 7.0.1.7
OS : SunOS 5.10
LogPrimaryFiles=3
LogSecondaryFiles=2
LogFilePages=1024
LogType=CIRCULAR
LogBufferPages=0
FDC files says the below
Probe Id :- XC338001
Major Errorcode :- xecE_W_UNEXPECTED_ASYNC_SIGNAL
| Probe Description :- AMQ6209: An unexpected asynchronous signal (2 : |
| SIGINT) has been received and ignored.
AND
| Probe Id :- XC307070
Major Errorcode :- xecL_W_LONG_LOCK_WAIT
Probe Description :- AMQ6150: WebSphere MQ semaphore is busy.
I have gone through some of the documents and posts given and they talk about the logging,curdepth,transactions etc but I dont see any problem in them.
Any clue to this is much appreciated as this is causing a problem of qmgr restart each time and that too in production. Also to note is that I am killing the processes of mq as endmqm -i is taking long time.
Thanks |
|
| Back to top |
|
 |
| bruce2359 |
| Posted: Fri Jul 27, 2012 10:04 am Post subject: Re: Queue Manager Goes for a Hang |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
| RaviKrG wrote: |
I have gone through some of the documents and posts given and they talk about the logging,curdepth,transactions etc but I dont see any problem in them.
|
Go to google. Type Probe Id XC307070. Click Search button.
There are several hits from IBM that suggest problem-determination steps, fixpacks, and so on. Did you follow these recommendations? What were the results?
_________________
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 |
|
|
| RaviKrG |
| Posted: Fri Jul 27, 2012 10:23 am Post subject: |
|
|
Master
Joined: 07 Sep 2008
Posts: 240
|
Yes.. I did ask Mr Google but could not get any thing which really resemble my problem... They talk about linear logging (but here it not linear), they talk about queue depths (the queue depths are not high).. they talk about increasing the LogBufferPages=0 to 4096 (max) but this is for Linear (here again we are not linear)..about fix pack we already have 7.0.1.7 and what I got is it should be more than 7.0.1.4
Now what more I can try as  as I don't see any use till now. May be I need some clue which I am not able to think |
|
| Back to top |
|
|
| RaviKrG |
| Posted: Fri Jul 27, 2012 10:27 am Post subject: |
|
|
Master
Joined: 07 Sep 2008
Posts: 240
|
|
| Back to top |
|
|
| mqjeff |
| Posted: Fri Jul 27, 2012 10:30 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
Did you confirm that the amount of log space you have assigned is sufficient for the transactions you are sending through?
Did you confirm that the unexpected async signal wasn't someone trying to telnet to the mq listener port?
Delete the queue manager, recreate it with 20 primary log files and 8 secondary. |
|
| Back to top |
|
|
| RaviKrG |
| Posted: Fri Jul 27, 2012 10:38 am Post subject: |
|
|
Master
Joined: 07 Sep 2008
Posts: 240
|
| Reg the file system .. that does not seems to a problem.. but I am thinking of increasing the Number of log files (P/S) and then check it. |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Fri Jul 27, 2012 11:01 am Post subject: Re: Queue Manager Goes for a Hang |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
| RaviKrG wrote: |
| Hi, we have a qmgr running on Solaris since very long time but now we are facing a problem which shows that queue manager is getting hung frequently. |
Since this qmgr has been a model of good behavior, what changed just prior to this bad behavior? New applications? Changed applications? Increased use of the applications?
_________________
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 |
|
|
| RaviKrG |
| Posted: Fri Jul 27, 2012 10:58 pm Post subject: |
|
|
Master
Joined: 07 Sep 2008
Posts: 240
|
| No change was done since very long but yes I did suspect the increase in load |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Sat Jul 28, 2012 4:48 am Post subject: Re: Queue Manager Goes for a Hang |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
| RaviKrG wrote: |
LogPrimaryFiles=3
LogSecondaryFiles=2
LogFilePages=1024
LogType=CIRCULAR
LogBufferPages=0 |
I believe that these are the mqs.ini initial values when you install WMQ. These are usually adequate for initial testing and validation of WMQ functionality, but not for production.
Please read the System administration manual (or InfoCenter equivalent) chapter titled Recovery and Problem Determination, which includes a section Calculating the size of the log.
_________________
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 |
|
|
| MBMQDeveloper |
| Posted: Wed Aug 01, 2012 2:07 am Post subject: |
|
|
Novice
Joined: 02 Jul 2012
Posts: 19
|
| Did you check the number of MQ processes running in the system? Such a scenario might occur if there is a faulty application accessing the Queue Manager leading to unusually high number of MQ processes. |
|
| Back to top |
|
|
| JasonE |
| Posted: Wed Aug 01, 2012 5:21 am Post subject: |
|
|
Grand Master
Joined: 03 Nov 2003
Posts: 1220
Location: Hursley
|
| Quote: |
| Probe Id :- XC307070, Major Errorcode :- xecL_W_LONG_LOCK_WAIT |
What is the call stack (first list of functions just under the big banner box).
Also, in general you get one of these fdc's when MQ has waited longer than a certain period for a lock, and then you should get another when the thing that held the lock for a long period of time finally releases it - do you have another from the release code, and if so, what was its call stack... if you dont, then deadlock is a possibility here.
| Quote: |
| Also to note is that I am killing the processes of mq as endmqm -i is taking long time. |
A long time, or never completing... If there's something deadlocked then it may never complete, but if its just slow, then killing it off feels like you are deferring pain for later on.
Do you have very deep queues? |
|
| Back to top |
|
|
|
|
