hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Yi Jin <y...@pivotal.io>
Subject Incomplete information of queue report when using FairScheduler
Date Wed, 16 Dec 2015 03:23:25 GMT
Hi all,

I am writing a program that uses YARN queue capacity data to estimate the
resource limits, i.e. as current application, how much resource can be
required at most, how much resource can be required at most without queue
overcommit.

For example, I configure one yarn queue named 'queue1' in scheduler
configuration file. Assume the application uses 20% resource of the whole
cluster now.

I find CapacityScheduler can provides queue1's runtime capacity, current
capacity and maximum capacity; While FairScheduler provides only queue1's
current capacity, and the current capacity value is set in field capacity.
For example,

Using CapacityScheduler, the report output looks like

QueueName: queue1
Capacity: 50%
CurrentCapacity:20%
MaximumCapacity:100%

Using FairScheduler, the report output looks like

QueueName: queue1
Capacity:20%  <------- in fact this value should be printed in the field
CurrentCapacity.
CurrentCapacity:0%
MaximumCapacity:0%

Is this a known issue ? Or is there a plan to improve FairScheduler to make
queue report well filled?

Any response is welcome, thank you all and looking forward to your response.

Best,
Yi

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message