dengzhhu653 commented on code in PR #5038:
URL: https://github.com/apache/hive/pull/5038#discussion_r1479896169
##########
standalone-metastore/metastore-server/src/main/java/org/apache/hadoop/hive/metastore/ObjectStore.java:
##########
@@ -5094,26 +5094,22 @@ public static String verifyStatsChangeCtx(String
fullTableName, Map<String, Stri
if (validWriteIds != null && writeId > 0) {
return null; // We have txn context.
}
- String oldVal = oldP == null ? null :
oldP.get(StatsSetupConst.COLUMN_STATS_ACCURATE);
- String newVal = newP == null ? null :
newP.get(StatsSetupConst.COLUMN_STATS_ACCURATE);
- // We don't need txn context is that stats state is not being changed.
- if (StringUtils.isEmpty(oldVal) && StringUtils.isEmpty(newVal)) {
+
+ if (!StatsSetupConst.areBasicStatsUptoDate(newP)) {
+ // The validWriteIds can be absent, for example, in case of Impala alter.
Review Comment:
As my understanding, there is no easy way to set `validWriteIds` to a
default value, and if the client is HiveServer2, we should always get a
non-null `validWriteIds` on the HMS side.
The main idea of the change is that as long as the stats is marked as
invalid(stale) on the new target table, we should just let the alter operation
go regardless of whether the `validWriteIds` is null or not, as it won't
pollute the Hive stats.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
