You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Roman Puchkovskiy (Jira)" <ji...@apache.org> on 2024/07/19 18:23:00 UTC

[jira] [Updated] (IGNITE-22786) Make 'slow message handling' threshold configurable

     [ https://issues.apache.org/jira/browse/IGNITE-22786?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Roman Puchkovskiy updated IGNITE-22786:
---------------------------------------
    Description: 
Logs like

[2024-07-19T19:10:28,498][WARN ][main][TrackableNetworkMessageHandler] Message handling has been too long [duration=88ms, message=[CmgInitMessageImpl [clusterName=cluster, cmgNodes=Set12 [ipact_n_0], initialClusterConfiguration=metaStorage\{idleSyncTimeInterval=10},replication\{idleSafeTimePropagationDuration=100},schemaSync\{delayDuration=100,maxClockSkew=7},security\{authentication{providers=[{name=default,type=basic,users=[{password=ignite,username=ignite}]}]}}, metaStorageNodes=Set12 [ipact_n_0]]]]

are currently written if message handling takes more than 5ms. This is reasonable for development environments, but in production such pauses could happen naturally, but such log messages would just scare the user.

We should either make the threshold configurable and make it a lot higher (like 500-1000ms) still leaving it 5ms in tests, or even disable this logging by default (as we have detection for inbound threads being 'stuck' for 500ms+.

> Make 'slow message handling' threshold configurable
> ---------------------------------------------------
>
>                 Key: IGNITE-22786
>                 URL: https://issues.apache.org/jira/browse/IGNITE-22786
>             Project: Ignite
>          Issue Type: Improvement
>            Reporter: Roman Puchkovskiy
>            Priority: Major
>              Labels: ignite-3, long-message-handling
>
> Logs like
> [2024-07-19T19:10:28,498][WARN ][main][TrackableNetworkMessageHandler] Message handling has been too long [duration=88ms, message=[CmgInitMessageImpl [clusterName=cluster, cmgNodes=Set12 [ipact_n_0], initialClusterConfiguration=metaStorage\{idleSyncTimeInterval=10},replication\{idleSafeTimePropagationDuration=100},schemaSync\{delayDuration=100,maxClockSkew=7},security\{authentication{providers=[{name=default,type=basic,users=[{password=ignite,username=ignite}]}]}}, metaStorageNodes=Set12 [ipact_n_0]]]]
> are currently written if message handling takes more than 5ms. This is reasonable for development environments, but in production such pauses could happen naturally, but such log messages would just scare the user.
> We should either make the threshold configurable and make it a lot higher (like 500-1000ms) still leaving it 5ms in tests, or even disable this logging by default (as we have detection for inbound threads being 'stuck' for 500ms+.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)