We are using Cloud Tasks to call an "on-prem" API gateway (using Http request). This API gateway (IBM API Connect) sits in front off an on-prem system (Oracle). This back end system can at times be very slow. >5s.
We are desperately trying to increase the throughput but “adjusting” the Cloud Task queue settings (like -max-dispatches-per-second etc).
gcloud tasks queues update queue-1 --max-dispatches-per-second=8 --max-concurrent-dispatches=16
But all we see when we “crank up” the Cloud Task settings is that yellow triangle telling us that we are “enforced" to lower rate due to "system resources".
My understanding is that the yellow triangle shows up due to “errors” from the API gateway we call. Basically GCP/Cloud Tasks re-acts "by it self" based on return codes/errors/time-outs/latency etc from the API end-point we are calling with the result of a very low rate/thru-put. Is this understanding correct? Can someone verify?
The GUI does say that "or because currently there is no instance available to execute a request". What instance are they talking about? So to me that means that there is a possibility that it's "GCP specific" resources that comes into the picture here and have an effect on the "enforced rate"? Or?
Anyway, any help/insight would be appreciated.
Thanks