question

AlejandroBenitezAragon-0465 avatar image
0 Votes"
AlejandroBenitezAragon-0465 asked ·

Azure Function Durable work-item count exceeds the configured maximum active work-item count

Hi guys, right now we are testing a durable functions thats use Fan out/fan in pattern. Everything is working well but when we try to scale the level of parallelism we get this message:

"Delaying work item fetching because the current active work-item count exceeds the configured maximum active work-item count (20)"

78477-imagen.png

We try to fix it setting this Application setting : WEBSITE_MAX_DYNAMIC_APPLICATION_SCALE_OUT = 200

78418-imagen.png

And this scale out parameter: Maximum Scale Out Limit : 200

78426-imagen.png

So guys, if you know the way to get the behavior we are looking for, will be nice if you could guide us.

Thanks for your help.




azure-functions
imagen.png (313.6 KiB)
imagen.png (5.0 KiB)
imagen.png (14.6 KiB)
10 |1000 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

1 Answer

PramodValavala-MSFT avatar image
1 Vote"
PramodValavala-MSFT answered ·

@AlejandroBenitezAragon-0465 The error that you are observing are at the host level (per instance) and are concurrency throttles that can be configured in host.json depending on your scenario.


· 2 ·
10 |1000 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

Thanks a lot.
Your answer has allowed me to move forward. Taking advantage of your knowledge, I want to ask:

  • When I execute the await starter.StartNewAsync and context.CallActivityAsync instructions in a serverless model, does azure scale virtual machines or does everything run within the same Azure Function instance?

Thx again.



0 Votes 0 ·
PramodValavala-MSFT avatar image PramodValavala-MSFT AlejandroBenitezAragon-0465 ·

@AlejandroBenitezAragon-0465 It would depend on the load your function app is under. Under no load, the same instance will run everything but as your function app scales out, activity functions could be picked up by any instance (from an internal queue).

The Performance and Scale should be give you a detailed overview of how durable functions handles this.


1 Vote 1 ·