Ok, if you need me to try anything in the server, let me know. I will be out from office until friday, but I still can connect and do some test.
Thanks again for your help! El miércoles, 13 de mayo de 2015, 15:37:38 (UTC-3), Niphlod escribió: > > there's a problem, I think. > max_empty_runs=5 means "do 5 loops, if you didn't pick or assign any > tasks, die". > you said your workers are started by a cron job every 2 minutes (god knows > why, but that's your choice). > your instantiation doesn't touch the heartbeat parameter, which is by > default 3 seconds. > this means that there is a "timeframe" of a taddle bit more than 15 > seconds to pick up "ready to process tasks". > Which, you said, are all there and ready to be processed. > A worker gets to rule out if he's an elegible "master" every 5 cycles (the > "ticker" who dispatches tasks). > The problem I'm about to highlight is subtle... when a worker needs to > rule out if he's an elegible master, he does also the "pruning" of dead > workers. > First of all, as soon as it's started, a worker does this thing (i.e. at > the very first loop). This means that if your workers are started every 2 > minutes, and there are ready to process tasks, at the very first round the > worker would prune "the last worker that died doing nothing" > Now, if the last worker didn't process any task and shutted down, he'd be > pruned as long as the last_heartbeat is 9 seconds (line 1015), because he'd > still be marked as an ACTIVE worker. > This doesn't add up with what you're experiencing... a worker with > max_empty_runs=5 started every 2 minutes would effectively have a worker > pick up the tasks as soon as it's started. > > This shouldn't happen...... > > DEBUG:web2py.scheduler.medios#27764:defining tables (migrate=False) > DEBUG:web2py.scheduler.medios#27764: freeing workers that have not > sent heartbeat > INFO:web2py.scheduler.medios#27764:nothing to do > DEBUG:web2py.scheduler.medios#27764:sleeping... > DEBUG:web2py.scheduler.medios#27764:empty runs 1/6 > INFO:web2py.scheduler.medios#27764:TICKER: I'm a ticker > DEBUG:web2py.scheduler.medios#27764:looping... > DEBUG:web2py.scheduler.medios#27764:Assigning tasks... > INFO:web2py.scheduler.medios#27764:TICKER: workers are 1 > INFO:web2py.scheduler.medios#27764:TICKER: tasks are 0 > > > tl;dr: glad that you sorted out, but I'm going to test things as they are > in your environment to try to reproduce the erratic behaviour. > -- Resources: - http://web2py.com - http://web2py.com/book (Documentation) - http://github.com/web2py/web2py (Source code) - https://code.google.com/p/web2py/issues/list (Report Issues) --- You received this message because you are subscribed to the Google Groups "web2py-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to web2py+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
