Uh oh!
There was an error while loading. Please reload this page.
TEZ-3363: Delete intermediate data at the vertex level for Shuffle Handler - #60
Conversation
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
shameersss1
commented
Feb 7, 2022
@abstractdog Could you please review the changes? |
@shameersss1: I'm more than interested in this patch, let me have some time to review it |
shameersss1
commented
Feb 8, 2022
@abstractdog - Thanks for showing interest to review. It has been pending for a while now. The high level idea behind this feature is that, Whenever all the dependent vertex of a particular vertex have succeeded we delete the vertex shuffle data of that particular/parent vertex. Testing Procedure
|
shameersss1
commented
Mar 2, 2022
@abstractdog Could you please review the changes? |
@shameersss1: sorry, I haven't had the chance, I want to test this on a cluster too, where I face some issues at the moment, also I'm busy with other changes, let me get back to you in 2 weeks, thanks for your patience! |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
This comment was marked as outdated.
This comment was marked as outdated.
abstractdog
commented
Mar 10, 2022
please rebase on top of master, latest compilation error could be due to TEZ-4227 |
shameersss1
commented
Mar 11, 2022
Rebased to latest master. pending tests |
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
Uh oh!
There was an error while loading. Please reload this page.
| vertex.appContext.getAppMaster().vertexComplete( | ||
| vertex.vertexId, nodes); | ||
| } else { | ||
| LOG.debug(String.format("The number of incomplete child vertex are %s for the vertex %s", |
There was a problem hiding this comment.
please use logger format here {} {}, String.format is not necessary
There was a problem hiding this comment.
ack. Will resolve in next revision.
Uh oh!
There was an error while loading. Please reload this page.
abstractdog
left a comment
There was a problem hiding this comment.
left very minor comments, this is very close @shameersss1 !
shameersss1
commented
Mar 13, 2022
I have resolved the comments. Thanks for your valuable feedback. |
tez-yetus
commented
Mar 13, 2022
🎊 +1 overall
This message was automatically generated. |
shameersss1
commented
Mar 16, 2022
@abstractdog - Are we good to take it forward? |
warnings here are easily addressable: https://ci-hadoop.apache.org/job/tez-multibranch/job/PR-60/7/artifact/out/diff-checkstyle-tez-plugins_tez-aux-services.txt |
| } | ||
| @VisibleForTesting | ||
| public VertexShuffleDataDeletionContext getVShuffleDeletionContext() { |
There was a problem hiding this comment.
this is not used currently in any of the tests, vShuffleDeletionContext can remain private to VertexImpl
I'm fine with leaving this part uncovered, can you please remove this method?
(sorry, last minute comments :) )
There was a problem hiding this comment.
This is used by testVertexShuffleDelete() in TestVertexImpl
There was a problem hiding this comment.
okay, missed it sorry, in this case, you can remove public keyword (make it package visible) to enhance "VisibleForTesting" behavior
There was a problem hiding this comment.
ack. I will fix it in next revision.
shameersss1
commented
Mar 16, 2022
ack. |
89c05dc to
db543c9Comparetez-yetus
commented
Mar 16, 2022
💔 -1 overall
This message was automatically generated. |
abstractdog
commented
Mar 16, 2022
can you check @shameersss1 if the latest conflict is because of TEZ-4359 and rebase if needed? |
This comment was marked as outdated.
This comment was marked as outdated.
tez-yetus
commented
Mar 16, 2022
🎊 +1 overall
This message was automatically generated. |
abstractdog
commented
Mar 16, 2022
merged to master, finally! thanks @shameersss1 for your tireless work and patience on this one! |
For applications like pig where processing times can be very long, applications may choose to delete intermediate data for a sub dag. For example if a DAG has synced data to HDFS, all upstream intermediate data can be safely deleted.