You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@atlas.apache.org by "Jayendra Parab (Jira)" <ji...@apache.org> on 2022/10/24 08:09:00 UTC
[jira] [Updated] (ATLAS-4691) Discrepancy in the atlas debug metrics between the active and the not active servers
[ https://issues.apache.org/jira/browse/ATLAS-4691?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jayendra Parab updated ATLAS-4691:
----------------------------------
Description:
There is a discrepancy in the atlas debug metrics between the active and the not active servers.
Steps to reproduce:
- Setup Atlas in HA mode with two instances of Atlas
- add "atlas.debug.metrics.enabled=true" to the Atlas configuration (conf/atlas-application.properties)
- restart Atlas
- open Atlas and create a new entity (e.g. hdfs_path)
- check the debug metrics on both atlas instances using API call: https://<atlas_host>:<atlas_port>/api/atlas/admin/debug/metrics
- The active instance returns correct metrics, whereas passive instance returns empty metrics
The reason for this is that the debug metrics are maintained in-memory in the active server. When "admin/debug/metrics" API call is accessed on the passive server, the request isn't redirected to the active server like it does for the other APIs, hence the passive node returns empty metrics.
was:
There is a discrepancy in the atlas debug metrics between the active and the not active servers.
Steps to reproduce:
- Setup Atlas in HA mode with two instances of Atlas
- add "atlas.debug.metrics.enabled=true" to the Atlas configuration (conf/atlas-application.properties_role_safety_valve)
- restart Atlas
- open Atlas and create a new entity (e.g. hdfs_path)
- check the debug metrics on both atlas instances using API call: https://<atlas_host>:<atlas_port>/api/atlas/admin/debug/metrics
- The active instance returns correct metrics, whereas passive instance returns empty metrics
The reason for this is that the debug metrics are maintained in-memory in the active server. When "admin/debug/metrics" API call is accessed on the passive server, the request isn't redirected to the active server like it does for the other APIs, hence the passive node returns empty metrics.
> Discrepancy in the atlas debug metrics between the active and the not active servers
> ------------------------------------------------------------------------------------
>
> Key: ATLAS-4691
> URL: https://issues.apache.org/jira/browse/ATLAS-4691
> Project: Atlas
> Issue Type: Bug
> Components: atlas-core
> Affects Versions: 2.2.0
> Reporter: Jayendra Parab
> Assignee: Jayendra Parab
> Priority: Major
>
> There is a discrepancy in the atlas debug metrics between the active and the not active servers.
> Steps to reproduce:
> - Setup Atlas in HA mode with two instances of Atlas
> - add "atlas.debug.metrics.enabled=true" to the Atlas configuration (conf/atlas-application.properties)
> - restart Atlas
> - open Atlas and create a new entity (e.g. hdfs_path)
> - check the debug metrics on both atlas instances using API call: https://<atlas_host>:<atlas_port>/api/atlas/admin/debug/metrics
> - The active instance returns correct metrics, whereas passive instance returns empty metrics
> The reason for this is that the debug metrics are maintained in-memory in the active server. When "admin/debug/metrics" API call is accessed on the passive server, the request isn't redirected to the active server like it does for the other APIs, hence the passive node returns empty metrics.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)