mesos / chronos

Fault tolerant job scheduler for Mesos which handles dependencies and ISO8601 based schedules
http://mesos.github.io/chronos/
Apache License 2.0
4.39k stars 529 forks source link

Run several tasks simultaneously on one slave #270

Open dendec opened 10 years ago

dendec commented 10 years ago

Good day, How to run several task simultaneously using Chronos on one mesos slave? I try to run second task when first task is still running, but Chronos doesn't start second task and mark it as "failure". In the log I see following messages:

Oct 01, 2014 11:44:44 AM com.airbnb.scheduler.state.MesosStatePersistenceStore com$airbnb$scheduler$state$MesosStatePersistenceStore$$remove INFO: Purging entry 'T_ct:1412178282735:0:2' via: org.apache.mesos.state.ZooKeeperState Oct 01, 2014 11:44:44 AM com.airbnb.scheduler.mesos.MesosJobFramework processTask INFO: Task 'ct:1412178282735:0:2' launched, status: 'DRIVER_RUNNING' Oct 01, 2014 11:44:44 AM com.airbnb.scheduler.mesos.MesosJobFramework getNextTask$1 INFO: No tasks scheduled! Declining offers Oct 01, 2014 11:44:44 AM com.airbnb.scheduler.mesos.MesosJobFramework statusUpdate WARNING: Task with id 'ct:1412178282735:0:2' LOST Oct 01, 2014 11:44:44 AM com.airbnb.scheduler.jobs.JobScheduler handleFailedTask WARNING: Task of job: 2 failed. Oct 01, 2014 11:44:44 AM com.airbnb.scheduler.jobs.JobScheduler handleFailedTask WARNING: Retrying job: 2, attempt: 0 Oct 01, 2014 11:44:44 AM com.airbnb.scheduler.state.MesosStatePersistenceStore persistData INFO: State T_ct:1412178282735:0:2 does not exist yet. Adding to state

As you see, second task is lost and failed. There is no such problem with using Marathon, it can run a lot of tasks simultaneously on Mesos. Do you know what can be reason of such behaviour?

Thank you, in advance.

P.S.: Our team find your project very useful for us, thank you for this work.

solarkennedy commented 9 years ago

Related to #532