CODEWITHSUNDEEP
hadoopmapreduceresourcemanager
Alvis
Alvis

Reputation: 119

What if the ResourceManager down?

In the newest version of Hadoop mapreduce(called 'Yarn'), JobTracker(exists in previous version) has been replaced by the ResourceManager(called 'RM') and ApplicationMaster. In official document about Yarn architecture, there are no words say that how many RMs are there in a MapReduce cluster, and the given graph about Yarn architecture shows only 1 RM exists in a cluster.

So, what if the only RM down? If there are several RMs, how do they work together?

Hope someone can explain it to me. Thanks.

Upvotes: 1

Views: 846

Answers (1)

Junayy
Junayy

Reputation: 1150

There is 1 RessourceManager per rack but you can have several racks in your cluster.

If you try to submit a job while RessourceManager is down, Hadoop will try to connect to the RessourceManager because it needs it to execute the job.

Here is an example of the logs when the RM is down and try to submit a job :

14/06/06 09:39:54 INFO ipc.Client: Retrying connect to server: hadoop01.sii.fr/10.6.6.211:8032. Already tried 1 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS)
14/06/06 09:39:55 INFO ipc.Client: Retrying connect to server: hadoop01.sii.fr/10.6.6.211:8032. Already tried 2 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS)
14/06/06 09:39:56 INFO ipc.Client: Retrying connect to server: hadoop01.sii.fr/10.6.6.211:8032. Already tried 3 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS)

When the RM is back, the job is submitting correctly.

Upvotes: 1

Related Questions