pvary commented on a change in pull request #2219:
URL: https://github.com/apache/hive/pull/2219#discussion_r620293681
##########
File path:
ql/src/java/org/apache/hadoop/hive/ql/ddl/table/AbstractAlterTableOperation.java
##########
@@ -138,8 +141,32 @@ private void finalizeAlterTableWithWriteIdOp(Table table,
Table oldTable, List<P
environmentContext.putToProperties(HiveMetaHook.ALTER_TABLE_OPERATION_TYPE,
desc.getType().name());
if (partitions == null) {
long writeId = desc.getWriteId() != null ? desc.getWriteId() : 0;
- context.getDb().alterTable(desc.getDbTableName(), table,
desc.isCascade(), environmentContext, true,
- writeId);
+ try {
+ context.getDb().alterTable(desc.getDbTableName(), table,
desc.isCascade(), environmentContext, true, writeId);
+ } catch (HiveException ex) {
+ if
(environmentContext.getProperties().containsKey(HiveMetaHook.INITIALIZE_ROLLBACK_ALTER)
&& Boolean
+
.valueOf(environmentContext.getProperties().get(HiveMetaHook.INITIALIZE_ROLLBACK_ALTER)))
{
+ // in case of rollback of alter table do the following:
+ // 1. drop the already altered table but keep the data files
+ // 2. recreate the original table
+ // 3. run msck repair to sync partitions on filesystem with
metastore
Review comment:
This could have unexpected result, if we have extra directories in the
table root, which are not added yet to the HMS. I think it would be good to
have a way to disable things
--
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.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
