[
https://issues.apache.org/jira/browse/IGNITE-26978?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Aleksandr Polovtsev updated IGNITE-26978:
-----------------------------------------
Description:
When working in Ignite 3 in real case scenarios, I found the current
information in logs to be quite lackluster. This is an umbrella ticket that
describes a bunch of small improvements to logs produced by some components.
# When a node starts, we should print existing zone and table configurations,
similar to what Ignite 2 does;
# There are messages related to long acquisitions of a
{{CheckpointReadWriteLock}} - I suggest to add some metrics to those messages,
like the number of readers, number of writers, queue size;
# "Non handled stable assignments for key" message - I suggest to add actual
assignments value, add more logs, e.g. reason for starting or stopping a
replica (similar to "Remove planned key because pending is empty" message);
There're also some annoying typos from the natural language perspective:
# "Remove planned key because pending is empty": "remove" should be "removed",
it is not obvious what "planned" and "pending" keys mean;
# "Checkpoint read lock acquisition has been timed out" should be "Checkpoint
read lock acquisition timeout";
# "Start building the index" should be "Started building index";
was:When working in Ignite 3 in real case scenarios, I found the current
information in logs to be quite lackluster. This is an umbrella ticket that
describes a bunch of small improvements to logs produced by some components.
> Logging improvements in Ignite 3
> --------------------------------
>
> Key: IGNITE-26978
> URL: https://issues.apache.org/jira/browse/IGNITE-26978
> Project: Ignite
> Issue Type: Epic
> Reporter: Aleksandr Polovtsev
> Priority: Major
> Labels: ignite-3
>
> When working in Ignite 3 in real case scenarios, I found the current
> information in logs to be quite lackluster. This is an umbrella ticket that
> describes a bunch of small improvements to logs produced by some components.
> # When a node starts, we should print existing zone and table configurations,
> similar to what Ignite 2 does;
> # There are messages related to long acquisitions of a
> {{CheckpointReadWriteLock}} - I suggest to add some metrics to those
> messages, like the number of readers, number of writers, queue size;
> # "Non handled stable assignments for key" message - I suggest to add actual
> assignments value, add more logs, e.g. reason for starting or stopping a
> replica (similar to "Remove planned key because pending is empty" message);
> There're also some annoying typos from the natural language perspective:
> # "Remove planned key because pending is empty": "remove" should be
> "removed", it is not obvious what "planned" and "pending" keys mean;
> # "Checkpoint read lock acquisition has been timed out" should be "Checkpoint
> read lock acquisition timeout";
> # "Start building the index" should be "Started building index";
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
