[
https://issues.apache.org/jira/browse/PHOENIX-4278?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16356329#comment-16356329
]
ASF GitHub Bot commented on PHOENIX-4278:
-----------------------------------------
Github user JamesRTaylor commented on a diff in the pull request:
https://github.com/apache/phoenix/pull/291#discussion_r166808341
--- Diff:
phoenix-core/src/main/java/org/apache/phoenix/query/ConnectionQueryServicesImpl.java
---
@@ -850,19 +849,12 @@ private void addCoprocessors(byte[] tableName,
HTableDescriptor descriptor, PTab
&& !SchemaUtil.isMetaTable(tableName)
&& !SchemaUtil.isStatsTable(tableName)) {
if (isTransactional) {
- if
(!descriptor.hasCoprocessor(PhoenixTransactionalIndexer.class.getName())) {
-
descriptor.addCoprocessor(PhoenixTransactionalIndexer.class.getName(), null,
priority, null);
- }
--- End diff --
This handles the case of the creation of new tables, but we'd need to
handle the case of existing tables as well. This is somewhat trickier because
ideally we'd want to handle the case in which the server jar has been upgraded,
but the client jar hasn't been. Until the client is upgraded (in which case
index maintenance will occur on the client side), we'd need to continue
performing index maintenance through this coprocessor. The somewhat clunky way
of doing that is to pass the client Phoenix version through the
PhoenixIndexMetaData. This would be somewhat painful.
Since we've labelled transactions as "beta", perhaps we can punt on this.
In this case, we'd require both the client and server jar to be upgraded at the
same time (if transactional tables are already being used). With this approach,
we'd want to modify PhoenixTransactionalIndexer to be an empty class derived
still from BaseRegionObserver.
> Implement pure client side transactional index maintenance
> ----------------------------------------------------------
>
> Key: PHOENIX-4278
> URL: https://issues.apache.org/jira/browse/PHOENIX-4278
> Project: Phoenix
> Issue Type: Improvement
> Reporter: James Taylor
> Assignee: Ohad Shacham
> Priority: Major
>
> The index maintenance for transactions follows the same model as non
> transactional tables - coprocessor based on data table updates that looks up
> previous row value to perform maintenance. This is necessary for non
> transactional tables to ensure the rows are locked so that a consistent view
> may be obtained. However, for transactional tables, the time stamp oracle
> ensures uniqueness of time stamps (via transaction IDs) and the filtering
> handles a scan seeing the "true" last committed value for a row. Thus,
> there's no hard dependency to perform this on the server side.
> Moving the index maintenance to the client side would prevent any RS->RS RPC
> calls (which have proved to be troublesome for HBase). It would require
> returning more data to the client (i.e. the prior row value), but this seems
> like a reasonable tradeoff.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
