HBASE-24524 SyncTable logging improvements#1878
Conversation
|
While troubleshooting mismatches in replication deployment, SyncTable logging can provide some insights on what is diverging between two clusters. One caveat, though, is that it logs diverging row key as hexdecimal values, which is not so useful for operators trying to figure out which rows are mismatching, ideally, this info should be human readable, so that operators could have the exact value they could use for querying the tables with other tools, such as hbase shell. Another issue is that while rows mismatches are logged as info, cell values mismatches are only logged as debug. In general, any of the mismatches would already be quite verbose, so ideally both should be logged in debug level. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
HorizonNet
left a comment
There was a problem hiding this comment.
LGTM. Is it worth calling out in a release note that the logging level has changed to prevent surprises by operators?
Makes sense to me, will do. |
Signed-off-by: Jan Hentschel <jan.hentschel@ultratendency.com> Signed-off-by: Viraj Jasani <vjasani@apache.org> (cherry picked from commit d8247eb)
Signed-off-by: Jan Hentschel <jan.hentschel@ultratendency.com> Signed-off-by: Viraj Jasani <vjasani@apache.org> (cherry picked from commit d8247eb)
Signed-off-by: Jan Hentschel <jan.hentschel@ultratendency.com> Signed-off-by: Viraj Jasani <vjasani@apache.org> (cherry picked from commit d8247eb)
Signed-off-by: Jan Hentschel <jan.hentschel@ultratendency.com> Signed-off-by: Viraj Jasani <vjasani@apache.org> (cherry picked from commit d8247eb)
Signed-off-by: Jan Hentschel <jan.hentschel@ultratendency.com> Signed-off-by: Viraj Jasani <vjasani@apache.org> (cherry picked from commit d8247eb)
No description provided.