Ashwani Raina has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/19413 )

Change subject: KUDU-3406 corrected estimate for ancient UNDO delta size
......................................................................


Patch Set 2: Code-Review+1

(4 comments)

Overall changes look good to me.

http://gerrit.cloudera.org:8080/#/c/19413/2//COMMIT_MSG
Commit Message:

http://gerrit.cloudera.org:8080/#/c/19413/2//COMMIT_MSG@16
PS2, Line 16: So, if a
            : delta file was lazily loaded without stats being read, 
DeltaTracker
            : assumed all its deltas were ancient.
I noticed this once during my tests. I got following trace during that test:
+++
LOG_WITH_PREFIX(INFO) << Substitute(                  
    "compaction isn't runnable because of too much data in "
    "ancient UNDO deltas: $0 out of $1 total bytes",
    ancient_undos_total_size, undos_total_size);
+++

I could not reproduce it later though.
Good that we found the root-cause.


http://gerrit.cloudera.org:8080/#/c/19413/2//COMMIT_MSG@19
PS2, Line 19: the
nit: for


http://gerrit.cloudera.org:8080/#/c/19413/2//COMMIT_MSG@25
PS2, Line 25: "metadata only"
Not a comment.

For my understanding, how does this type of workload help in manifesting the 
issue?


http://gerrit.cloudera.org:8080/#/c/19413/2/src/kudu/tablet/rowset.h
File src/kudu/tablet/rowset.h:

http://gerrit.cloudera.org:8080/#/c/19413/2/src/kudu/tablet/rowset.h@115
PS2, Line 115: OVERESTIMATE,
             :     UNDERESTIMATE,
nit: A one liner comment to explain in what case, one would choose this type.
e.g. For CompactRowSetsOp, why it would be UNDERESTIMATE and for 
UndoDeltaBlockGCOp, why an OVERESTIMATE.



--
To view, visit http://gerrit.cloudera.org:8080/19413
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I17bddae86f84792caf14fb1e11a6e1c0d7a92b56
Gerrit-Change-Number: 19413
Gerrit-PatchSet: 2
Gerrit-Owner: Alexey Serbin <[email protected]>
Gerrit-Reviewer: Ashwani Raina <[email protected]>
Gerrit-Reviewer: Kudu Jenkins (120)
Gerrit-Comment-Date: Thu, 12 Jan 2023 07:22:46 +0000
Gerrit-HasComments: Yes

Reply via email to