This is an automated email from the ASF dual-hosted git repository.
mgreber pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/kudu.git
The following commit(s) were added to refs/heads/master by this push:
new 513862e73 [docs] Document tablet_history_max_age_sec behavior
513862e73 is described below
commit 513862e737d5a68c5683eadd1efd28dacc2dbd74
Author: Abhishek Chennaka <[email protected]>
AuthorDate: Wed Nov 19 14:35:37 2025 -0800
[docs] Document tablet_history_max_age_sec behavior
This updates the known issues section with the behavior of the flag
--tablet_history_max_age_sec and suggestions to improve cluster
performance.
Change-Id: Ic86db57eab6da21c0fe2e8721f120318bdccba79
Reviewed-on: http://gerrit.cloudera.org:8080/23694
Tested-by: Kudu Jenkins
Reviewed-by: Ashwani Raina <[email protected]>
Reviewed-by: Alexey Serbin <[email protected]>
---
docs/known_issues.adoc | 11 +++++++++++
1 file changed, 11 insertions(+)
diff --git a/docs/known_issues.adoc b/docs/known_issues.adoc
index 2bdcf0786..d58ac51ad 100644
--- a/docs/known_issues.adoc
+++ b/docs/known_issues.adoc
@@ -217,3 +217,14 @@ to communicate only the most important known issues.
If using SSSD in secure Kudu cluster deployments, make sure SSSD packages are
of version 1.16 or newer
(see link:https://issues.apache.org/jira/browse/KUDU-3217[KUDU-3217]).
+
+* The default value of --tablet_history_max_age_sec was increased from 15
minutes
+ to 1 week in Kudu 1.10. While maintaining a longer history window can be
useful
+ for features such as incremental Kudu backups and time-travel queries, it
also
+ comes with significant resource costs such as higher memory usage, increased
CPU load,
+ higher rate of filesystem read/write operations and more open file
descriptors due to
+ storing all table data changes in delta files. Scanning rows that are
updated frequently
+ imposes additional overhead, since all the accumulated deltas must be
processed, negatively
+ affecting performance. Decreasing the value of this flag can improve
performance, and if
+ the workload does not rely on long history retention, it is safe to reduce
the value
+ (see http://issues.apache.org/jira/browse/KUDU-3583[KUDU-3583]).
