Author |
Message
|
bduncan |
Posted: Wed Aug 01, 2001 3:29 pm Post subject: |
|
|
Padawan
Joined: 11 Apr 2001 Posts: 1554 Location: Silicon Valley
|
We are currently running Candle to monitor things like channel status and queue depth. The alerts are passed on to Tivoli, and we receive pages describing the alert. At least that's how it's supposed to work... We have been getting a lot of false pages, and multiple pages for the same event. In other words, we go over queue depth high, and get paged. Our interval is set to 10 minutes, and every 10 minutes we get paged about the queue depth high even though the depth may have gone below queue depth high. We also get pages about channels being in doubt and when we go and look in the queue manager they are running fine. Candle has admitted that there is a bug causing the repeating pages; essentially Candle advised us to check the box called "restart policy" for our queue depth high event. When we didn't have it checked, we would get a single queue depth high page, and then never get paged again even if a different queue on a different queue manager went over queue depth high. This is because the queue depth high event was assigned to all the queue managers in our system. But now that we've checked that box, we get multiple pages for the same event on the same queue. Is anyone else having behavior like this with Candle????
_________________ Brandon Duncan
IBM Certified MQSeries Specialist
MQSeries.net forum moderator |
|
Back to top |
|
|
blamk |
Posted: Thu Aug 30, 2001 9:39 am Post subject: |
|
|
Newbie
Joined: 29 Aug 2001 Posts: 1
|
I would wonder whether the depth of the queue in question had been reduced below the "Low Depth Threshold", which would turn off the High Depth Threshold alert and should thus cause the policy to stop being activated and the repeated pages to cease.
Barry Lamkin
Candle Corporation |
|
Back to top |
|
|
bduncan |
Posted: Thu Aug 30, 2001 12:23 pm Post subject: |
|
|
Padawan
Joined: 11 Apr 2001 Posts: 1554 Location: Silicon Valley
|
I see your point, but my question is this. If I am asking Candle to page me when a queue depth high event occurs, I would expect the pages to stop when the current queue depth is no longer above the queue depth high threshold, not when it goes all the way down to queue depth low. I am simply asking Candle to check a true/false condition: is the current depth above queue depth high? Once this ceases to be true my pages have to stop.
_________________ Brandon Duncan
IBM Certified MQSeries Specialist
MQSeries.net forum moderator |
|
Back to top |
|
|
RogerLacroix |
Posted: Sun Sep 02, 2001 10:06 pm Post subject: |
|
|
Jedi Knight
Joined: 15 May 2001 Posts: 3258 Location: London, ON Canada
|
Hi all,
There are probably 10 ways to deal with this but the easy way that I can think of, is to use a policy that is actived when the queue depth high situation is fired (true).
In the policy have it check current depth against max depth. If they are not equal then reset the situation of queue depth high to false (pages will stop).
later
Roger...
|
|
Back to top |
|
|
bduncan |
Posted: Mon Sep 03, 2001 10:25 am Post subject: |
|
|
Padawan
Joined: 11 Apr 2001 Posts: 1554 Location: Silicon Valley
|
I assume that in your statement:
"In the policy have it check current depth against max depth"
You meant to say "depth specified by queue depth high" not "max depth" right? If we do it the way it is specified, then we won't get any pages until the queue actually fills up, right?
_________________ Brandon Duncan
IBM Certified MQSeries Specialist
MQSeries.net forum moderator |
|
Back to top |
|
|
RogerLacroix |
Posted: Mon Sep 03, 2001 9:34 pm Post subject: |
|
|
Jedi Knight
Joined: 15 May 2001 Posts: 3258 Location: London, ON Canada
|
True. When I wrote my reply, I was thinking of the queue full event rather than queue high threshold event.
later
Roger...
|
|
Back to top |
|
|
cvshiva |
Posted: Sun Mar 24, 2002 2:15 am Post subject: |
|
|
Apprentice
Joined: 04 Mar 2002 Posts: 35 Location: Chennai
|
Hi Brandon,
We do use Candle monitoring for all our production systems in the bank. But we don't have paging,we do monitor all the events you mentioned , but never the alerts behaved falsely.Particularly talking abt Channels IN DOUBT.. Channels will be running when in doubt, but the MCA is confused whether to Commit or Backout the last batch of messages due to some interruption in the communication between MCA's ( Eg. Network breakdown when the Acknowledgement from Rcvr to Sdr is on the cable , and sdr never recd acknowledgement for the message it sent). This needs manual intervention.. You have to Resolve the channel with action commit or backout.
RESOLVE CHANNEL( ) ACTION(COMMIT|BACKOUT)..
To know whether the channel is in doubt
DIS CHS(*) INDOUBT
None of the INDOUBT(XXX) displayed should be (YES) ..
If you use MQSeries Explorer , please customise the columns displayed for channels ( in the right pane of MQSeries Explorer Window) to show Indoubt status also.
_________________ Ramnath Shiva
IBM Certified SOA Specialist
IBM Certified MQSeries Specialist
Standard Scope International Pvt Ltd , Chennai |
|
Back to top |
|
|
blane99 |
Posted: Wed Jun 12, 2002 5:40 am Post subject: |
|
|
Apprentice
Joined: 12 Jun 2002 Posts: 41
|
Have you tried a situation rather than policy? We are sending email for same via Action in a situaton and have no issues. WHat version of Candle?
Re: indoubts - we experienced the same (especially on OS/390) and found increasing the interval for the situtaion from 1 to 2 or 3 did the trick. |
|
Back to top |
|
|
|