Problem with PegaRpaService
We have a connection problem beetween Pega Robot Manager and Virtual Machine.
Robot is online for about 3.5 h after starting PegaRpaService on VM. Connection is terminated after this time.If we restart RpaService the problem repeats.
Error from RpaService attached.
Pegasystems Inc.
GB
Is this happening for all VM or just for this one? do you see any errors in Robot Manager? are you using a proxy or a load balancer? if so, can you see any errors there?
Uniqa Tu SA
PL
We are testing now, so we have only one VM. We are using proxy, and we don't see any other errors in RM, but maybe we're not looking where we should be.
I have attached versions of our Pega applications, if that helps anything
David Taliga
AT
Hey Pawel, from our experence. Pega RPA Service should be runnning 24/7 if VM is online. Its very unusual that RPAService stops. Maybe you can check if Pega Platform server is not restarted while RPAService is running.
Uniqa Tu SA
PL
Hi David,
Pega Platform is online, VM is online, and PegaRpaService is running. Everything looks normal.
For about 3.5 hours everything works fine and then an error occurs. And we don't know why?
Updated: 9 Mar 2023 11:14 EST
David Taliga
AT
Hey Pawel,
Your error message says:
"Request header is too large" Could not it be connected to memory issue ? are not you filling Pega with heartbeat requests and Pega is not able to receive it any further because of memory ?
I am sorry I have not other explanation.
Updated: 13 Mar 2023 3:58 EDT
Uniqa Tu SA
PL
The default value is 30 seconds, and that value is set.
I checked other intervals and they also have default values.
But do you know what request is sent by RpaService? And what is the header?
Because I see only these 4 lines from RpaService log:
-Sending heartbeat. -XXXXX - Is Robot running? 'False'. -Heartbeat request robot count '1'. -NetworkClient - sending POST request to URL 'http://XXXXXX:8080/prweb/api/robotics/v2/heartbeat?type=serviceoperator'.
Maybe there is some place where we can check what request is actually sent?
Uniqa Tu SA
PL
We found the logs we were looking for. They are located here: Records->Integration-Resources->Service Package->robotics->heartbeat
We enabled service monitoring and now we know that each request is longer, because "accept" part of it increases:
HTTP/1.1 POST http://XXXXX/prweb/api/robotics/v2/heartbeat?type=serviceoperator authorization:**masked** content-length:87 expect:100-continue cookie:**masked** host:XXXX content-type:application/json; charset=utf-8 accept:application/json, application/json, application/json, application/json.......
After a few hours "application/json " is repeated dozens of times. Should it work like this?