[
https://issues.apache.org/jira/browse/HIVE-17884?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16217943#comment-16217943
]
Prasanth Jayachandran commented on HIVE-17884:
----------------------------------------------
Also would recommend one other minor change (either in this patch of followup
jira), while creating resource plan will be better to remove '_' from query
parallelism
QUERY_PARALLELISM -> QUERY PARALLELISM
we can use '_' for trigger expression counter names.
Also, trigger expression counters does not accept values other than integer.
Something like below will throw error
WHEN HDFS_BYTES_READ > 10GB DO KILL
from usability perspective, it will be easier to specify 10GB vs values in
bytes. Similarly for time based counters, WHEN EXECUTION_TIME > 2 hours DO
KILL. ExpressionFactory.java can parse such counters ("10 GB").
> Implement create, alter and drop workload management triggers.
> --------------------------------------------------------------
>
> Key: HIVE-17884
> URL: https://issues.apache.org/jira/browse/HIVE-17884
> Project: Hive
> Issue Type: Sub-task
> Reporter: Harish Jaiprakash
> Assignee: Harish Jaiprakash
> Attachments: HIVE-17884.01.patch
>
>
> Implement triggers for workload management:
> The commands to be implemented:
> CREATE TRIGGER `resourceplan_name`.`trigger_name` WHEN condition DO action;
> condition is a boolean expression: variable operator value types with 'AND'
> and 'OR' support.
> action is currently: KILL or MOVE TO pool;
> ALTER TRIGGER `plan_name`.`trigger_name` WHEN condition DO action;
> DROP TRIGGER `plan_name`.`trigger_name`;
> Also add WM_TRIGGERS to information schema.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
