| Author |
Message
|
| steve_ghb |
 Posted: Tue Aug 01, 2006 9:25 am Post subject: workitems in suspending state |
 |
|
Novice
Joined: 15 Jun 2005
Posts: 13
|
We have hundreds of process instances that are in suspending state. We have a script that is supposed to terminate any process instances older than a specific date but it doesn't seem to terminate these.
Could a large number of these be causing performance problems we are having? Even if not, how can we get rid of them without doing it one by one? I tried terminating one of them and a got a timeout so that option may not be available anyway.
Thanks |
|
| Back to top |
|
 |
| sarasu |
| Posted: Tue Aug 01, 2006 9:46 am Post subject: |
|
|
Master
Joined: 02 Feb 2006
Posts: 229
|
| If you are using normal workflow runtime client, use a fileter to sort out all the instances and select all and delete will delete all the process instances. You can do that using API aswell. |
|
| Back to top |
|
|
| koko |
| Posted: Tue Aug 01, 2006 10:02 am Post subject: |
|
|
Master
Joined: 26 Sep 2003
Posts: 206
|
| Quote: |
| We have hundreds of process instances that are in suspending state |
Do you know why this happened?
_________________
Thanks
Koko |
|
| Back to top |
|
|
| steve_ghb |
| Posted: Tue Aug 01, 2006 10:14 am Post subject: |
|
|
Novice
Joined: 15 Jun 2005
Posts: 13
|
| Quote: |
| Do you know why this happened? |
Users are supposed to suspend a workitem if they cannot process it right away, for whatever reason sometimes they sit in suspending state instead of becoming suspended. Haven't been able to figure out why. |
|
| Back to top |
|
|
| koko |
| Posted: Tue Aug 01, 2006 10:26 am Post subject: |
|
|
Master
Joined: 26 Sep 2003
Posts: 206
|
| Quote: |
| Users are supposed to suspend a workitem if they cannot process it right away, for whatever reason sometimes they sit in suspending state instead of becoming suspended |
Initally you said that process instances are in suspending state not workitems.
_________________
Thanks
Koko |
|
| Back to top |
|
|
| steve_ghb |
| Posted: Tue Aug 01, 2006 10:47 am Post subject: |
|
|
Novice
Joined: 15 Jun 2005
Posts: 13
|
| Oops, I meant suspend the instance |
|
| Back to top |
|
|
| steve_ghb |
| Posted: Tue Aug 01, 2006 12:50 pm Post subject: |
|
|
Novice
Joined: 15 Jun 2005
Posts: 13
|
Which state value from process_inst table is the one for 'suspending'?
Actually if anyone can point me to a list of what all the state values translate to that would be great.
Thanks alot |
|
| Back to top |
|
|
| jmac |
| Posted: Tue Aug 01, 2006 3:25 pm Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
Here are the values for ProcessInstanceState
| Code: |
enum FmcjProcessInstanceStateValue
{
Fmc_PS_NotSet = 0,
Fmc_PS_Ready = 1, Fmc_PS_Running = 2, Fmc_PS_Finished = 4,
Fmc_PS_Terminated = 8, Fmc_PS_Suspended = 16,
Fmc_PS_Terminating = 32, Fmc_PS_Suspending = 64,
Fmc_PS_Deleted = 128
}; |
You can find the values by doing a search on the *.h files int he API subdirectory of your MQWF installation
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
| xiaojiahuochen |
| Posted: Tue Aug 01, 2006 9:08 pm Post subject: |
|
|
Newbie
Joined: 11 Jul 2006
Posts: 6
|
first use query filter to query processinstance list then
user code
ProcessInstance.terminate()
can terminate |
|
| Back to top |
|
|
| hos |
| Posted: Tue Aug 01, 2006 11:48 pm Post subject: |
|
|
Chevalier
Joined: 03 Feb 2002
Posts: 470
|
Hi,
have a look into the Programming Guide, chapter 'Process Instance Actions'. There you'll find a state transition diagram that shows you the PI sates, the available transitions and the conditions for a state transition.
Furthermore the Suspend() API describes the conditions for a state transition:
"The process instance remains in state Suspending as long as there are running program activity implementations, suspending non-autonomous subprocesses, or checked-out work items. When the activity implementations completed their executions and the non-autonomous subprocesses reached the Suspended state, and when the checked-out work items are checked in, the process instance is put into the Suspended state."
I personally do not think that it is a good idea to give end-users the authority to suspend process instances. |
|
| Back to top |
|
|
| steve_ghb |
| Posted: Wed Aug 02, 2006 5:23 am Post subject: |
|
|
Novice
Joined: 15 Jun 2005
Posts: 13
|
| Thanks all for the information, I'll have to do some research into this too, especially about why these instances are in suspending state. I can't imagine there are that many workitems still checked out for instances that are several years old. There must be something still running like you said. |
|
| Back to top |
|
|
| steve_ghb |
| Posted: Wed Aug 02, 2006 11:42 am Post subject: |
|
|
Novice
Joined: 15 Jun 2005
Posts: 13
|
Problem is bigger than I thought at first. Not so much worried about the suspending instances we can clean those out at some point.
Queried the db this morning and got the following numbers:
# of process instances: 617,810 (I didn't filter for top level only)
# in state=4(finished): 352,300
Looks like our scheduling server isn't moving instances from finished to deleted state. Keep time is 1 day but I know for sure we didn't finish over 350,000 instances in the last day or two.
Here's our settings:
- Scheduling server settings
- Check interval PT1M
- Start mode Immediate
- Notification check interval PT1M
- Create notification items threshold 10
- Delete notification items threshold 100
- Cleanup server settings
- Check interval PT1H
- Start mode Deferred
- Server start time Everyday at 02:00
- Server stop time 04:30
- Cleanup time P1D
- Cleanup delay time PT1H
Any ideas why these instances aren't getting changed to state 128? also does anyone know what the ibm recommended settings are for these servers? I don't know whay they were set to 'out of the box' and I'm sure they've been changed.
Thanks |
|
| Back to top |
|
|
| jmac |
| Posted: Wed Aug 02, 2006 12:24 pm Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
Are you certain that the cleanup server is running?
Do you see the cleanup server started/stopped messages in the log
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
| steve_ghb |
| Posted: Wed Aug 02, 2006 1:08 pm Post subject: |
|
|
Novice
Joined: 15 Jun 2005
Posts: 13
|
I thought the cleanup server physically deletes instances that are in state 128, and the scheduling server changes finished processes to state 128 for the cleanup server to delete?
Anyway, we ran queries this morning before, during, and after the cleanup server was scheduled (I haven't checked the logs yet) and here's what we got:
before: 12,325 in state 128
during: 12,325 in state 128 (don't know why yet it didn't do anything at first)
after: 9,570 in state 128
Definitely looks like there's a problem with the cleanup server but I'm more concerned that over half the process instances are finished and not getting cleaned up.
I ran another query and over 175,000 of the finished instances have a completed time of before July 21. They are supposed to be cleaned up after 1 day. |
|
| Back to top |
|
|
| Ratan |
| Posted: Wed Aug 02, 2006 1:19 pm Post subject: |
|
|
Grand Master
Joined: 18 Jul 2002
Posts: 1245
|
| Quote: |
| I ran another query and over 175,000 of the finished instances have a completed time of before July 21. They are supposed to be cleaned up after 1 day. |
Just wondering if you might be looking at finished Activity Instances. May be the process instance is not finished yet.
_________________
-Ratan |
|
| Back to top |
|
|
|
|
