Uh oh!
There was an error while loading. Please reload this page.
HBASE-28055 Performance improvement for scan over several stores. - #5379
Conversation
ss77892
commented
Aug 31, 2023
No new UT is required. The one for HBASE-19863 works just fine with the problem. |
Apache-HBase
commented
Aug 31, 2023
💔 -1 overall
This message was automatically generated. |
ss77892
commented
Aug 31, 2023
retest |
Apache-HBase
commented
Aug 31, 2023
💔 -1 overall
This message was automatically generated. |
ss77892
commented
Aug 31, 2023
Test failure is not related to the patch and it passed locally. |
Apache-HBase
commented
Aug 31, 2023
🎊 +1 overall
This message was automatically generated. |
Apache-HBase
commented
Aug 31, 2023
🎊 +1 overall
This message was automatically generated. |
Apache-HBase
commented
Aug 31, 2023
🎊 +1 overall
This message was automatically generated. |
Apache-HBase
commented
Aug 31, 2023
💔 -1 overall
This message was automatically generated. |
ankitsinghal
left a comment
There was a problem hiding this comment.
As per Sergey explanation in private chat (copying below) , it seems cell.getTimestamp() == PrivateConstants.OLDEST_TIMESTAMP is enough to regard the cell as fake
There was an assumption that compareKeyForNextColumn would tell us whether the nextCell lower than the current cell. But the result of compareKeyForNextColumn is based on the fact of availability of the next column. So, to get the expected result the regular comparison of cell and nextCell should be used. At the same time the situation when nextCell is lower than the current cell happens only when the current cell is artificial (fake) one, so the 'official' way to do that is to check the timestamp
taklwu
left a comment
There was a problem hiding this comment.
LGTM, basically this change shortcut part of the function of matcher.compareKeyForNextColumn and compare the LastOnRowCell with the HConstants.OLDEST_TIMESTAMP before we can skip to the next column
Apache-HBase
commented
Sep 5, 2023
🎊 +1 overall
This message was automatically generated. |
Apache-HBase
commented
Sep 5, 2023
🎊 +1 overall
This message was automatically generated. |
Apache-HBase
commented
Sep 5, 2023
🎊 +1 overall
This message was automatically generated. |
ndimiduk
commented
Sep 11, 2023
…res. (apache#5379)" This reverts commit 4c1fd0d.
GeorryHuang
commented
Sep 12, 2023
taklwu
commented
Sep 12, 2023
My bad , I will create PRs for backport next time |
…ache#5379) Signed-off-by: Ankit Singhal <ankit@apache.org> Signed-off-by: Tak Lon (Stephen) Wu <taklwu@apache.org>
No description provided.