Hi all! Help me track this behavior. I have an job (datomic-plugin/read-log) and all working fine, until the system remain several hours idle, after that the job still up if i curl http://localhost:8080/replica/jobs but the job dont read more from DB if i insert something, only back to work if i change tenancy ID, sync my database and submit the job again. I lost something? There is any properties to sleep the job or similar?

That should definitely still work. Is there anything interesting (exceptions?) in the onyx.log?

No exceptions, if i turn on info mode the log keep logging this

17-08-07 18:17:15 ip-10-0-1-50 INFO [onyx.peer.task-lifecycle:215] - Checkpointed output c514efa8-faca-f1bc-f3e8-c78bdf5d2a92 30 137 :out 0 :output
17-08-07 18:17:18 ip-10-0-1-50 INFO [onyx.peer.task-lifecycle:177] - Checkpointed input c514efa8-faca-f1bc-f3e8-c78bdf5d2a92 30 138 :read-log 0 :input
17-08-07 18:17:18 ip-10-0-1-50 INFO [onyx.peer.task-lifecycle:215] - Checkpointed output c514efa8-faca-f1bc-f3e8-c78bdf5d2a92 30 138 :out-websocket 0 :output
17-08-07 18:17:18 ip-10-0-1-50 INFO [onyx.peer.task-lifecycle:215] - Checkpointed output c514efa8-faca-f1bc-f3e8-c78bdf5d2a92 30 138 :out-email 0 :output
17-08-07 18:17:18 ip-10-0-1-50 INFO [onyx.peer.task-lifecycle:215] - Checkpointed output c514efa8-faca-f1bc-f3e8-c78bdf5d2a92 30 138 :out 0 :output

Looks like the job is still chugging along as expected then. Hmm.

yeah, but my input spy function never call

Yeah, I just mean that what you said about the job being up is true. I’m not sure why it’s not continuing to read.

Do you think can be any kind of transactor and datomic-plugin idle disconnect?

Possibly, but probably not what’s going on. Is your spy fn on the input task? That’d help reduce the chances that it’s problems with the segments flowing downstream.

yes, it is in my :onyx/plugin :onyx.plugin.datomic/read-log

K. I assume you’re using 0.10?

yeap

I can give you a version with some logging to help you debug.

would be nice!

just a min

@lellis please try 0.10.0.1-debug1 with :onyx/batch-timeout increased (to reduce logging), and logging :info level

batch-timeout of 1000 or 2000 would probably be OK

nice ty!