[
https://issues.apache.org/jira/browse/PHOENIX-1198?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14134812#comment-14134812
]
James Taylor commented on PHOENIX-1198:
---------------------------------------
Nice work, [~elilevine]. I made one minor suggestion that I think would help
usability.
bq. 2. Another approach is to add annotations to Mutations and Scans that are
passed into Phoenix coprocessor methods using
OperationWithAttributes.setAttribute() method. Should be fine for Scans but is
probably overkill to add to every Mutation if a coprocessor method accepts
Mutation batches. Maybe just add to the first Mutation in a batch?
Except for DDL operations, all our coprocessors pass information through
attributes as you've described, so I think this is our best bet (see
BaseScannerRegionObserver for list of attributes used for Scans). For updates
to tables with indexes, we have a mechanism in place already which we could
likely generalize a bit for the table with no indexes case too. If it's a
single Put in the batch, we pass the attributes in the Put. Otherwise, we pass
it through an extra once per region server endpoint coprocessor call in the
ServerCache (see IndexMetaDataCacheClient.addIndexMetadataCache()).
> Add ability to pass custom tags to be added to log lines
> --------------------------------------------------------
>
> Key: PHOENIX-1198
> URL: https://issues.apache.org/jira/browse/PHOENIX-1198
> Project: Phoenix
> Issue Type: Improvement
> Reporter: Eli Levine
> Assignee: Eli Levine
> Fix For: 5.0.0, 4.1
>
>
> These tags can be passed in either when creating connections or calling
> upsert/select. Similar to PHOENIX-1196. Maybe they can shared the same
> mechanism for passing in values to be logged/traced.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
