You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-dev@hadoop.apache.org by "Haibo Chen (JIRA)" <ji...@apache.org> on 2018/03/17 18:03:00 UTC
[jira] [Created] (YARN-8042) Improve debugging on ATSv2 reader
server
Haibo Chen created YARN-8042:
--------------------------------
Summary: Improve debugging on ATSv2 reader server
Key: YARN-8042
URL: https://issues.apache.org/jira/browse/YARN-8042
Project: Hadoop YARN
Issue Type: Improvement
Affects Versions: 3.0.0
Reporter: Haibo Chen
It's been inconvenient to debug issues that happened on the read path. Typically, a query sent from a client is parsed into a TimelineReaderContext, TimelineEntityFilters and TimelineDataToRetrieve which are independent of the underlying backend storage implementations. Then the general ATSv2 building blocks are then translated into a SCAN and GET query in HBase with specified row keys and filters.
To facilitate easy debugging, additional debug level logging messages (ideally that can be enabled dynamically without restarting TimelineReaderServer process) can be added at the boundary to narrow down the scope of investigation.
A good example of this is logging the scan or get query before it is sent to HBase and the result after the query returns from HBase. YARN support folks who are not necessarily HBase experts can present the debug messages to HBase experts and get help. (I had to remotely connect to TimelineReaderServer, set up breakpoints and get the hbase queries every time I suspect there is a bug in HBase)
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-dev-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-dev-help@hadoop.apache.org