Vihang Karajgaonkar has posted comments on this change. ( http://gerrit.cloudera.org:8080/13300 )
Change subject: IMPALA-8447: [DOCS] INSERT event is supported in automatic invalidation ...................................................................... Patch Set 1: (10 comments) http://gerrit.cloudera.org:8080/#/c/13300/1/docs/topics/impala_metadata.xml File docs/topics/impala_metadata.xml: http://gerrit.cloudera.org:8080/#/c/13300/1/docs/topics/impala_metadata.xml@135 PS1, Line 135: notifications should it be "notification events"? http://gerrit.cloudera.org:8080/#/c/13300/1/docs/topics/impala_metadata.xml@146 PS1, Line 146: <codeph>ADD</codeph>, or <codeph>DROP</codeph> their : partitions After IMPALA-7973, the add, drop and alter partition events trigger a refresh table action. Alter table is still invalidates the table http://gerrit.cloudera.org:8080/#/c/13300/1/docs/topics/impala_metadata.xml@161 PS1, Line 161: Refreshes I think may be we should also mention that if the table is not loaded, the event processor does not refresh the table. This is done to avoid loading the table unnecessarily in the catalog in case the table is unlikely to be used in the future. http://gerrit.cloudera.org:8080/#/c/13300/1/docs/topics/impala_metadata.xml@198 PS1, Line 198: non-zero may be say positive value to be accurate. Non-zero implies negative values are okay too. http://gerrit.cloudera.org:8080/#/c/13300/1/docs/topics/impala_metadata.xml@208 PS1, Line 208: such as adding new data to existing : tables/partitions from Spark I think this is a very generic statement which may cause users to avoid using this feature. We should try to be more specific on exact cases where it may not work when using spark. We can do this in a separate patch if you like. http://gerrit.cloudera.org:8080/#/c/13300/1/docs/topics/impala_metadata.xml@221 PS1, Line 221: dump data directly I think a better way to say this would be ".. and add or remove data into table by adding files directly on the filesystem". We should also recommend using Hive's LOAD DATA command to do the data load in such cases, so that event processor can act on the events generated by the LOAD command. http://gerrit.cloudera.org:8080/#/c/13300/1/docs/topics/impala_metadata.xml@233 PS1, Line 233: Inserts into tables from Hive are ignored. This can be misinterpreted as insert from Hive is not supported. I think insert from hive works from the end-user's perspective. Whether it works because of the underlying invalidate or refresh is a implementation detail which they will most likely not care about. So may be just skip this line? http://gerrit.cloudera.org:8080/#/c/13300/1/docs/topics/impala_metadata.xml@237 PS1, Line 237: invalidate invalidated. Also, I think this is a implementation detail which do not need to be documented. http://gerrit.cloudera.org:8080/#/c/13300/1/docs/topics/impala_metadata.xml@242 PS1, Line 242: Unlike other types of events, inserts by Impala triggers table refresh This can be removed. I think its implementation detail and users may not get any additional value by knowing this. http://gerrit.cloudera.org:8080/#/c/13300/1/docs/topics/impala_metadata.xml@267 PS1, Line 267: org.apache.hive.hcatalog.listener.DbNotificationListener After IMPALA-7971, we also need hive.metastore.dml.events set to true -- To view, visit http://gerrit.cloudera.org:8080/13300 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I68133b0beeb15cacc73829b8a8b0838fc7f4b7d8 Gerrit-Change-Number: 13300 Gerrit-PatchSet: 1 Gerrit-Owner: Alex Rodoni <arod...@cloudera.com> Gerrit-Reviewer: Alex Rodoni <arod...@cloudera.com> Gerrit-Reviewer: Bharath Krishna <bhar...@cloudera.com> Gerrit-Reviewer: Impala Public Jenkins <impala-public-jenk...@cloudera.com> Gerrit-Reviewer: Vihang Karajgaonkar <vih...@cloudera.com> Gerrit-Comment-Date: Tue, 14 May 2019 19:18:31 +0000 Gerrit-HasComments: Yes
