[
https://issues.apache.org/jira/browse/HBASE-30473?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Junegunn Choi updated HBASE-30473:
----------------------------------
Description:
h2. Problem
Operators can disable cluster-wide services through HBase shell switches, but
most switch states are not exported as metrics. For example,
{{cleaner_chore_switch false}} disables HFile and WAL cleaning, and operators
must run {{cleaner_chore_enabled}} to check that state. A monitoring system
cannot directly detect that the cleaner was left disabled. (Only the balancer
exports {{tag.isBalancerActive}} as a string tag in a separate metrics source.)
h2. Proposed Fix
Expose nine numeric gauges under
{{Hadoop:service=HBase,name=Master,sub=Server}}:
- {{balancerEnabled}}
- {{normalizerEnabled}}
- {{splitEnabled}}
- {{mergeEnabled}}
- {{cleanerChoreEnabled}}
- {{catalogJanitorEnabled}}
- {{snapshotCleanupEnabled}}
- {{replicationPeerModificationEnabled}} (2.6+)
- {{rpcThrottleEnabled}} (2.6+)
Use {{1}} for enabled and {{0}} for disabled. {{rpcThrottleEnabled}} reports
{{-1}} when quota support is enabled but the quota manager is not ready.
was:
h2. Problem
Operators can disable cluster-wide services through HBase shell switches, but
most switch states are not exported as metrics. For example,
{{cleaner_chore_switch false}} disables HFile and WAL cleaning, and operators
must run {{cleaner_chore_enabled}} to check that state. A monitoring system
cannot directly detect that the cleaner was left disabled. (Only the balancer
exports {{tag.isBalancerActive}} as a string tag in a separate metrics source.)
h2. Proposed Fix
Expose nine numeric gauges under
{{Hadoop:service=HBase,name=Master,sub=Server}}:
- {{balancerEnabled}}
- {{normalizerEnabled}}
- {{splitEnabled}}
- {{mergeEnabled}}
- {{cleanerChoreEnabled}}
- {{catalogJanitorEnabled}}
- {{snapshotCleanupEnabled}}
- {{replicationPeerModificationEnabled}}
- {{rpcThrottleEnabled}}
Use {{1}} for enabled and {{0}} for disabled. {{rpcThrottleEnabled}} reports
{{-1}} when quota support is enabled but the quota manager is not ready.
> Expose Master switch states as numeric metrics
> ----------------------------------------------
>
> Key: HBASE-30473
> URL: https://issues.apache.org/jira/browse/HBASE-30473
> Project: HBase
> Issue Type: Improvement
> Components: master, metrics
> Reporter: Junegunn Choi
> Assignee: Junegunn Choi
> Priority: Minor
> Labels: pull-request-available
> Fix For: 4.0.0-alpha-1, 2.7.0, 3.1.0, 3.0.1, 2.6.8, 2.5.17
>
>
> h2. Problem
> Operators can disable cluster-wide services through HBase shell switches, but
> most switch states are not exported as metrics. For example,
> {{cleaner_chore_switch false}} disables HFile and WAL cleaning, and operators
> must run {{cleaner_chore_enabled}} to check that state. A monitoring system
> cannot directly detect that the cleaner was left disabled. (Only the balancer
> exports {{tag.isBalancerActive}} as a string tag in a separate metrics
> source.)
> h2. Proposed Fix
> Expose nine numeric gauges under
> {{Hadoop:service=HBase,name=Master,sub=Server}}:
> - {{balancerEnabled}}
> - {{normalizerEnabled}}
> - {{splitEnabled}}
> - {{mergeEnabled}}
> - {{cleanerChoreEnabled}}
> - {{catalogJanitorEnabled}}
> - {{snapshotCleanupEnabled}}
> - {{replicationPeerModificationEnabled}} (2.6+)
> - {{rpcThrottleEnabled}} (2.6+)
> Use {{1}} for enabled and {{0}} for disabled. {{rpcThrottleEnabled}} reports
> {{-1}} when quota support is enabled but the quota manager is not ready.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
