hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andrew Purtell (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-12938) Upgrade HTrace to a recent supportable incubating version
Date Wed, 28 Jan 2015 19:27:34 GMT

    [ https://issues.apache.org/jira/browse/HBASE-12938?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14295672#comment-14295672

Andrew Purtell commented on HBASE-12938:

Ah, I see what you mean. The HBase span receiver isn't contained within our sources, it's
in HTrace. 

A rolling upgrade is expected to replace all local jars including dependencies. What cross-server
issues do you see [~busbey] ? Did HTrace make incompatible wire format changes? 

I'm not sure we have much of a choice anyway, the HTrace situation in 0.98 isn't supportable.
I'd hate for folks to have to make private patches to local HBase forks for usable tracing.

> Upgrade HTrace to a recent supportable incubating version
> ---------------------------------------------------------
>                 Key: HBASE-12938
>                 URL: https://issues.apache.org/jira/browse/HBASE-12938
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Andrew Purtell
>             Fix For: 0.98.11
> In 0.98 we have an old htrace (still using the org.cloudera.htrace package) and since
the introduction of htrace code, htrace itself first moved to org.htrace, then became an incubating
project. I filed this as a bug because the HTrace version we reference in 0.98 is of little
to no use going forward. Unfortunately we must make a disruptive change, although it looks
to be mostly fixing up imports, we expose no HTrace classes to HBase configuration, and where
we extend HTrace classes in our code, those HBase classes are in hbase-server and not tagged
for public consumption.   

This message was sent by Atlassian JIRA

View raw message