[
https://issues.apache.org/jira/browse/IGNITE-17352?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Konstantin Orlov updated IGNITE-17352:
--------------------------------------
Description:
To make the logs consistent we need to revise the code of module raft.
Here are some rule to follow:
* Use the following message template: <Short description of the event> [<key
value pairs describing context of the event>]. Example:
** LOG.info("Start new raft node [group={}, peers={}]", groupId, peers)
** LOG.debug("Replicator added [node={}, term={}, peer={}]", getNodeId(),
this.currTerm, peer);
* Prefer simple syntax constructions.
** (x) New leader was elected for the raft group={} of partition={}, table={}
and pending reconfiguration to peers={} was discovered
(/) New leader elected. Going to reconfigure peers [group={}, partition={},
table={}, peers={}]
** (x) Table with name '{}' has been created in schema with name '{}'
(/) Table created [schemaName={}, tableName={}]
* Avoid using a dot at the end – this is a log statement, not an essay
* Avoid using ERROR level just because we are going to log an exception.
Please remember ERROR is an actionable level, which means that attention of an
operation team is required. Besides, an ERROR should require an _immediate_
action. The action required should be clear from a message.
* WARN is an actionable level as well, but the action could be postponed. Use
this level to notify an operation team about upcoming failures (i.e. disk usage
is above certain threshold) or conditions that may affect the cluster, but the
resolution of such condition is not urgent (like happened page replacement;
this is debatable)
* INFO is a regular level to log any unfrequent event happened in the cluster,
like join of a new node
* DEBUG and TRACE levels don't have any particular rules, but it's nice to
have log message consistent across all levels
was:
To make the logs consistent we need to revise the code of module raft.
Here are some rule to follow:
* Use the following message template: <Short description of the event> [<key
value pairs describing context of the event>]. Example:
** LOG.info("Start new raft node [group={}, peers={}]", groupId, peers)
** LOG.debug("Replicator added [node={}, term={}, peer={}]", getNodeId(),
this.currTerm, peer);
* Prefer simple syntax constructions.
** (x) New leader was elected for the raft group={} of partition={}, table={}
and pending reconfiguration to peers={} was discovered
(/) New leader elected. Going to reconfigure peers [group={}, partition={},
table={}, peers={}]
** (x) Table with name '{}' has been created in schema with name '{}'
(/) Table created [schemaName={}, tableName={}]
* Avoid using a dot at the end – this is a log statement, not an essay
* Avoid using ERROR level just because we are going to log an exception.
Please remember ERROR is an actionable level, which means that attention of an
operation team is required. Besides, an ERROR should require an _immediate_
action. The action required should be clear from a message.
* WARN is an actionable level as well, but the action could be postponed. Use
this level to notify an operation team about upcoming failures (i.e. disk usage
is above certain threshold) or conditions that may affect the cluster, but the
resolution of such condition is not urgent (like happened page replacement, but
this is debatable)
* INFO is a regular level to log any unfrequent event happened in the cluster,
like join of a new node
* DEBUG and TRACE levels don't have any particular rules, but it's nice to
have log message consistent across all levels
> Revise module RAFT to align logs with proposed guidelines
> ---------------------------------------------------------
>
> Key: IGNITE-17352
> URL: https://issues.apache.org/jira/browse/IGNITE-17352
> Project: Ignite
> Issue Type: Improvement
> Reporter: Konstantin Orlov
> Priority: Major
> Labels: ignite-3
>
> To make the logs consistent we need to revise the code of module raft.
> Here are some rule to follow:
> * Use the following message template: <Short description of the event> [<key
> value pairs describing context of the event>]. Example:
> ** LOG.info("Start new raft node [group={}, peers={}]", groupId, peers)
> ** LOG.debug("Replicator added [node={}, term={}, peer={}]", getNodeId(),
> this.currTerm, peer);
> * Prefer simple syntax constructions.
> ** (x) New leader was elected for the raft group={} of partition={},
> table={} and pending reconfiguration to peers={} was discovered
> (/) New leader elected. Going to reconfigure peers [group={}, partition={},
> table={}, peers={}]
> ** (x) Table with name '{}' has been created in schema with name '{}'
> (/) Table created [schemaName={}, tableName={}]
> * Avoid using a dot at the end – this is a log statement, not an essay
> * Avoid using ERROR level just because we are going to log an exception.
> Please remember ERROR is an actionable level, which means that attention of
> an operation team is required. Besides, an ERROR should require an
> _immediate_ action. The action required should be clear from a message.
> * WARN is an actionable level as well, but the action could be postponed.
> Use this level to notify an operation team about upcoming failures (i.e. disk
> usage is above certain threshold) or conditions that may affect the cluster,
> but the resolution of such condition is not urgent (like happened page
> replacement; this is debatable)
> * INFO is a regular level to log any unfrequent event happened in the
> cluster, like join of a new node
> * DEBUG and TRACE levels don't have any particular rules, but it's nice to
> have log message consistent across all levels
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
