I wanted to bring to your attention a new issue I am experiencing with OpenRPA. Currently, it seems that OpenRPA is stuck in connections, which shows that there are sometimes more than three connections open at once. If the internet connection is interrupted or the server shuts down due to electricity shortages, this problem persists. As you can see from the attached screenshot, there are three connections still open for a period of time, even while other workflows are running. I would appreciate your assistance in resolving this issue as soon as possible.
Hey
That is not a connection counter. That is the number of packages that has not been sent.
Most packages ( like sending a message queue message, or insert/updating something in the database ) gets saved in memory and re-send when connection is re-established.
Hmm interesting, so why were these packages didn’t re-established connection when the server is back online? does this lead to eating more memory? I believe it’s still waiting to establish a connection and the server is alive.
I suggest giving a period of time (timeout) for each package to re-establish the connection. then retry it till the server is back online.
Once the server is back online, if the package is valid (not expired) then let’s re-establish the connection otherwise drop the connection in order to save memory.
That is not how it works, so would require a complete rewrite, I’m not interested in that right now.
Also I don’t see why each package should have a connection, that would completely overload openflow, if every single package made a connection. The whole point of websockets is to keep a single connection alive so both client and server can send messages when they have something interesting to say.
I agree, unsent packages should probably not be saved for ever, but on the other hand, if a robot is running for so long without having connection to openflow, then it’s probably better to just run it in offline mode anyway, so not sure how big of a problem this is.
Ram usage is not related to that. Not with only 4 messages in queue.
Have you updated openrpa? I fixed a serious memory leak in latest version.
If yes, then feel free to create and upload a memory dump ( Not to this forum ) then i can have a look at what is consuming the memory
