[
https://issues.apache.org/jira/browse/OOZIE-1737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13933531#comment-13933531
]
Robert Kanter commented on OOZIE-1737:
--------------------------------------
I'm not sure how this would work exactly, but would it make sense to give each
job its own log file? That way, when streaming, we don't have to do any
searching/parsing -- we'd just stream that job's log file as-is.
> Oozie log streaming is slow
> ---------------------------
>
> Key: OOZIE-1737
> URL: https://issues.apache.org/jira/browse/OOZIE-1737
> Project: Oozie
> Issue Type: Bug
> Reporter: purshotam shah
>
> Most of cases, Oozie search huge set of log files and return lot of logs.
> Most of our users has complained that log streaming is slow and they can't
> easily find relevant information.
> To make log streaming fast and precise we should provide few options (which
> user can specify for log streaming) like
> 1. Start time : Start date (if number, then offset(in hour) from
> coord/wf/bundle start)
> 2. End time : End date (if number, then offset from coord/wf/bundle start)
> 3. Log level :
> Setting log level to ERROR, will only give erred log and exception.
> In most of cases user want to find out error message and exception.
> 4. Number of lines : limit number of lines of logs from start date.
> 5. Recent logs: In most of the case user is only interested in recent
> activity.
> Similar to https://issues.apache.org/jira/browse/OOZIE-1716.
> User should be able to specify recent logs in hours.
> 6. Grep for pattern.
> + in addition to this, we also need to clean up log statements( which may
> also include consolidating log statements).
--
This message was sent by Atlassian JIRA
(v6.2#6252)
