I’ve got 10 Integration Servers in WM8.2 connected to one single broker.
Some days ago, we’ve noticed some triggers were disconnected and that the broker run-time data was full and that IS were storing documents locally.
After identifying the queue with all documents, we’ve restarted 2 faulty IS and the documents started to flow again.
However, after all queues have been emptied, it looks like the broker run-time data is still at 22% (ie. not empty…).
On another hand, for one single instance of these Integration Servers, the outbound store has got 27000 documents and is not actually sending anything to the broker.
However, connection to broker is marked as OK.
Based on this page: http://webmethodswiki.blogspot.fr/2015/12/webmethods-broker-failure-publishing.html, my understanding is that the IS outbound store does not free up because the Broker Data utilization is above 10%.
I cannot tell how true this is since it’s hard to understand why other Integration Servers actually emptied their local outbound store under these conditions though this specific one didn’t.
So, in a nutshell:
- Broker server Run-time data is at 22%
- All queues on all broker instances are empty (or so MWS says)
- One single integration server out of 10 does not drain its outbound store though it says it’s properly connected to the broker.
Does anyone have any suggestion regarding:
- Why the Broker Server run-time data is not empty though the queues seem to be empty ?
- How I can free up the Broker Run-time data ?
- How I can make the Integration Server to drain its local outbound store ?
Thanks in advance for helping.
please check the following:
In the Clients Tab of Messaging UI go to Options and check the “Show All” check box.
Go back to Search and klick Search.
Then sort the table by Capacity.
You will find some internal queues containing documents.
I.e. SyncClient, which is used for syncing doc types between Brokers.
When you are only using one Broker this queue can be emptied without problems.
What do you mean by Broker instances?
Is your Broker using splitted data files (one for config data and one for the throughput data)?
What is yout exact FixLevel?
Can you try to restart the Broker Server while keeping the ISes running?
I’ve already done so: there are really no queue with any document in it.
I sort by the “Docs Queued” criteria and all I can see is 0 everywhere.
What I call Broker instance is what you see listed in Messaging > Broker Servers > Brokers .
I call Broker Server what you see in Messaging > Broker Servers > Servers and for a single “Broker Server” you can have multiple Brokers (what I call Broker Instances) created in it (Messaging > Broker Servers > Brokers).
This is just to make a difference as some people call “Broker Server” what’s in “Servers” and what’s in “Brokers”, which would be quite confusing.
We’ve got actually 2 separate files for Configuration and Run-time data.
Configuration file is pretty small as it is less than 7MB.
Run-time file is 230MB even with emptied queues.
Broker version is 18.104.22.168.259 100511
I can’t really start the Broker Server at the moment as it happens there are some on-going tests at the moment for our flows.
can you provide a ScreenShot of the output of the Broker Server details → Utilization?
Looks like you are running plain GA version 8.2 SP2 without Fixes.
There is a Broker 8.2 SP3 available for which a BR_8.2_SP3_Core_Fix16 exists.
Please provide output of UpdateManager (View installed Fixes), at least for the Broker-related components.