Reply
New Contributor
Posts: 2
Registered: ‎09-07-2015

Doubts on few topics

Hi

 

Would you someone please provides answers / links for below questions?

 

  1. Why we need to update parameter values in CM only, why not in XML files itself?
  2. My Hive job is taking long time, what might be the possible problems?, what are all thing i need to look into?
  3. Where default reducer run?
  4. For particular user I want to revoke web server access , how I can achieve this
  5. Can capacity and fair scheduler in same cluster possible, how? (for user1-Fair, for User2-Cpacity)

 

Thnaks in advance.

 

Kavitha.

Posts: 1,903
Kudos: 435
Solutions: 307
Registered: ‎07-31-2013

Re: Doubts on few topics

> Why we need to update parameter values in CM only, why not in XML files itself?

CM is a configuration management layer for your cluster. A good read to fully understand CM and all the benefits it offers is here: http://www.cloudera.com/content/cloudera/en/documentation/core/latest/topics/cm_intro_primer.html

> My Hive job is taking long time, what might be the possible problems?, what are all thing i need to look into?

Start with the client/HS2 log to identify what phase of the query is it slow during (and also if a job is slow, or if a local task is). Subsequently, check up counters and skew probabilities if its a straggling task slowing the wider job. These are just some ideas.

> Where default reducer run?

What is a 'default reducer'? Reducers are scheduled randomly across the cluster, as they have no real local data dependency. There's no 'default' location for them if that's what you mean to ask.

> For particular user I want to revoke web server access , how I can achieve this

Two-step process: Enable authentication on web consoles, and then use the ACL configs for Web UIs to only allow specific users.

> Can capacity and fair scheduler in same cluster possible, how? (for user1-Fair, for User2-Cpacity)

No - The RM does not support running multiple scheduler implementations within the same runtime.