Hi, due to a bug we have in an application one of our tasks is increa

moritzw over 10 years ago

Hi,

due to a bug we have in an application one of our tasks is increasing by one every day. Because of that the process will run into an error once the MAX_NODE_INSTANCE property is reached. The processes can in theory run for many years. Our workaround for this is to set the property to infinity in custom.properties:
server.conf.processcommon.MAX_NODE_INSTANCES=2147483647

Anyone aware of any risks we're taking with this? There are currently only about 6-7 instances that have the problem. We've adresses the bug in a current hotfix, so there will be no future processes with that problem.

TIA, Moritz...

Discussion posts and replies are publicly visible

0 hemalathat
A Score Level 1
over 10 years ago

Its not best practice to change the property server.conf.processcommon.MAX_NODE_INSTANCES in Custom.property file. So you can consider changing your design. Given is the link for your reference: forum.appian.com/.../e-63749
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 moritzw over 10 years ago

Thanks for the reply. I realize that this is not best practice. We just have to change it because of the bug we currently have. Fortunately, there are just a few instances affected.

I am trying to figure out why it's not best practice. What kind of impact can have a larger number? I am guessing it's maily because of memory usage issues that arise with this?

Thanks, Moritz
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Myles Weber
Appian Employee
over 10 years ago

From what I recall, there are 2 main reasons for this setting.
a) It is to help avoid poor designs as there are better performing alternative ways
b) It avoids unintentional mistakes of run-away processes
There could be performance implications, but without testing that pattern, I couldn't say for sure.
If it is just once a day for many years, I'd suggest using a lower value than near infinity as you still get some of the above benefits.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 moritzw over 10 years ago

Thanks! That makes sense. I'll put it on 5000. That allows us to run the process for more than 10 years before it errors out. This should be fine.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 marky over 7 years ago

I would suggest looking the in other tab of the node and selecting the property to delete previous completed/cancelled instances. This is something that can be done to fix the future instances, and can be done as a process instance repair activity in production as well. This might fix your problem and avoid changing the property.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel