[GOBBLIN-711] stopping the master in cluster mode fails the job

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: gobblin-cluster
• Assignee: Hung Tran
• Reporter: Jay Sen
• Created: 23/Mar/19 23:37
• Updated: 03/Apr/19 05:25

Description

stopping the master process in cluster mode has some issues

1. graceful shutdown takes more than 40 seconds when the job is in progress ( no worker node is actually executing, master process is only tracking the job status here )
2. wrongfully reports the error ( org.apache.gobblin.runtime.AbstractJobLauncher  - Failed to launch and run job job_PullFromWikipedia_1553650384618) also ( org.apache.gobblin.cluster.HelixRetriggeringJobCallable  - Failed to run job PullFromWikipedia)
3. wrongfully reports job is done ( org.apache.gobblin.cluster.GobblinHelixJobLauncher  - Job job_PullFromWikipedia_1553650384618 is done, remove from running map.)

attached (master-stop.log) shows the full log of stopping and starting master and screenshots.