hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Appy (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HBASE-15809) Basic Replication WebUI
Date Thu, 04 Jan 2018 02:28:01 GMT

     [ https://issues.apache.org/jira/browse/HBASE-15809?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Appy updated HBASE-15809:
    Attachment: rep_web_ui.zip

Attached rep_web_ui.zip containing frontend code (html and js).
The UI is specifically designed for scale i.e. ~500-1000 nodes cluster. At that scale, it's
impossible to put numbers/raw data in top level UI. It's rather colorful :)
Files description:
*.csv : Fake data used for designing the Ui. Assuming cluster size of 500nodes.
*.min.js: Standard js libs
index.html: The code i wrote to generate UI from fake data. This is the main part. Contains
js too.

What's missing is backend to actually collect the data and feed into UI. We had few general
ideas, but none seemed trivial. Let me put the ones i remember here:
# UI uses ajax to query RegionServers' jmx every X sec. X depends on cluster size and will
likely be in order of minutes for 1000 node cluster. Might lead to interesting challenges
with browser. Idk.
# Each RS sends selective replication metrics to master, and UI fetches everything in one
sweep. Since data can be ~MBs for large cluster, can put non-negligible load on master. Refresh
rate of UI should be set accrodingly.

> Basic Replication WebUI
> -----------------------
>                 Key: HBASE-15809
>                 URL: https://issues.apache.org/jira/browse/HBASE-15809
>             Project: HBase
>          Issue Type: New Feature
>          Components: Replication, UI
>    Affects Versions: 2.0.0
>            Reporter: Matteo Bertozzi
>            Assignee: Appy
>            Priority: Minor
>             Fix For: 2.0.0, 1.5.0
>         Attachments: HBASE-15809-v0.patch, HBASE-15809-v0.png, HBASE-15809-v1.patch,
> At the moment the only way to have some insight on replication from the webui is looking
at zkdump and metrics.
> the basic information useful to get started debugging are: peer information and the view
of WALs offsets for each peer.
> https://reviews.apache.org/r/47275/

This message was sent by Atlassian JIRA

View raw message