[
https://issues.apache.org/jira/browse/IGNITE-26179?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Vladimir Steshin updated IGNITE-26179:
--------------------------------------
Description:
Among the settings of _TcpCommunication_ we have parameters like
{_}usePairedConnections, connectionsPerNode{_}, {_}messageQueueLimit{_}.
These settings can improve throughput, reduce latency and limit unsent data
memory consumption . But we have no indicator hinting that the current settings
of the pool aren't enough.
We've met user experience where a user had to use _connectionsPerNode_
parameter. But to find it out, an external analysis tool (like JRF) was
required.
To provide information of connection pool utilization we may expose some
metrics. Various common connection pools often supply its metrics.
The metrics might be:
* Pool size
* Number of currently open connections
* Number of active connections
* Average connection idle time
* Average connection lifetime
* Number of closed connections
And also
* Number of pending messages to each node
was:
Among the settings of _TcpCommunication_ we have parameters like
{_}usePairedConnections, connectionsPerNode{_}, {_}messageQueueLimit{_}.
These settings can improve throughput, reduce latency and limit unsent data
memory consumption . But we have no indicator hinting that the current settings
of the pool aren't enough.
We've met user experience where a user had to use _connectionsPerNode_
parameter. But to find it out, an external analysis tool (like JRF) was
required.
To provide information of connection pool utilization we may expose some
metrics. Various common connection pools often supply its metrics.
The metrics might be:
* Pool size
* Number of currently open connections
* Number of active connections
* Average connection idle time
* Average connection lifetime
* Number of closed connections
And also
* Number of unsent messages to each node
> Metrics of TCPCommunication connection pool.
> --------------------------------------------
>
> Key: IGNITE-26179
> URL: https://issues.apache.org/jira/browse/IGNITE-26179
> Project: Ignite
> Issue Type: Improvement
> Reporter: Vladimir Steshin
> Assignee: Vladimir Steshin
> Priority: Major
> Labels: ise
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Among the settings of _TcpCommunication_ we have parameters like
> {_}usePairedConnections, connectionsPerNode{_}, {_}messageQueueLimit{_}.
> These settings can improve throughput, reduce latency and limit unsent data
> memory consumption . But we have no indicator hinting that the current
> settings of the pool aren't enough.
> We've met user experience where a user had to use _connectionsPerNode_
> parameter. But to find it out, an external analysis tool (like JRF) was
> required.
> To provide information of connection pool utilization we may expose some
> metrics. Various common connection pools often supply its metrics.
> The metrics might be:
> * Pool size
> * Number of currently open connections
> * Number of active connections
> * Average connection idle time
> * Average connection lifetime
> * Number of closed connections
> And also
> * Number of pending messages to each node
--
This message was sent by Atlassian Jira
(v8.20.10#820010)