Hi
We have an on premise Octopus server and thousands of external clients configured with a polling tentacle. It is not possible to use listening mode due to client firewall configurations.
We are facing an issue with excessive income internet traffic to the point where we have to actually block the polling.
We want to adjust the polling interval to at least 2 minutes to reduce the load…
We can adjust the server timeouts to accommodate the increased interval, but I cannot find out how to adjust the tentacle polling interval
The nearest help document that I can find relates to a 2016 version saying it is not currently possible. I cannot believe that this would not have been addressed in 4 years.
Unfortunately, this feature hasn’t been added yet and isn’t currently on our roadmap for any future releases.
I’ll notify our products team that there is still interest in this feature, however, as it has been several years since it was first requested I would assume that there is a reason for it not being implemented.
Hi Paul,
Thanks for getting back to me. It seems to be very unusual that this is not something that has gained interest. The polling tentacle interval is extreme by default (every 5 seconds).
You can imagine the load that puts on an internet connection with thousands of clients.
While each poll in itself is not large, when you combine that with the shear volume of 10,000+ polls per minute it floods our connection to the point that nothing works.
Our firewall detects this as a DDOS attack and blocks them, meaning that the product is unusable.
Regards
Ian
The tentacles wouldn’t be connecting that often. Each tentacle should be initiating a new connection with the server at most every 30s.
The Tentacle opens a request to Octopus to get the next message. The server delays a response until either there is a message or PollingQueueWaitTimeout (30s default) is up. It then returns the message or null. The Tentacle will then immediately ask for the next message.