I’m running a Crossbar instance in AWS (specifically, a t2.medium). I’ve noticed that the host keeps eating up more and more memory until it starts getting dangerously close to the memory available on the machine. After restarting the host, the same issue repeats itself. You can see a Grafana graph over the last couple of weeks below:
The dips on the above graph are dates on which I restarted the instance. On 10/13, I upgraded Crossbar from 17.6.1.post3 to the latest version 17.9.2. Crossbar is basically the only thing running on the machine. Additionally, when I run
top, my top hit is the crossbar-worker:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
1325 crossbar 20 0 1357344 1.215g 11212 S 3.0 31.5 262:40.67 crossbar-worker
So, it seems clear that Crossbar is chewing up more memory every day. I created a session and called “wamp.registration.list” and confirmed that our registrations are where I’d expect them to be (only about 160). I also checked the “wamp.registration.get”, “wamp.registration.list_callees”, and “wamp.registration.count_callees” and everything looks reasonable for each registration.
Finally, I tried connecting with Pyrasite, but was unable to get it to connect to the process.
Is anyone familiar with something I could be doing that might cause Crossbar to chew up extra memory? Does anyone have any other suggestions for how I might debug this issue?