Re: Redis usage

Steven Zou

Please see comments inline.





Steven Zou(邹佳)


Engineer, MAP(Modern Application Platform), VMware R&D | Harbor Maintainer


Mail: szou@...


Cell: +8618600021252

Addr: 9F Tower C, Raycom Info Tech Park, No. 2 Kexueyuan South Road Haidian District, Beijing 100738 China




From: <harbor-users@...> on behalf of "bruno via Lists.Cncf.Io" <>
Reply-To: "harbor-users@..." <harbor-users@...>
Date: Tuesday, March 31, 2020 at 15:11
To: "harbor-users@..." <harbor-users@...>
Subject: [harbor-users] Redis usage


[Edited Message Follows]
[Reason: add clarification on why I sent this msg]

Many open-source project use redis just as a cache.
What about harbor? why it use it?


>>Steven: harbor core and registry components are using Redis as cache.  Jobservice is using Redis as job queue that supports retry and failover restore functions.

I did already my homework on this one, and my last step is too look at harbor code itself... but I have a terrible experiences with beego

I want to be sure if redis data must be:

- persitent on redis statefulset upgrade

>>Steven: Y

- is that data important and must be backup? but as PostgreSQL is already there... I assume critical data that need integrity is stored there. I assume it might just be user sessions.

>>Steven: The underlying job framework is using Redis as queue and some cron spec of periodical jobs launched by jobservice are also stored in Redis

- is redis can be rebuild from PostgreSQL, just not efficient when redis start from a blank PVC.

>>Steven: So far, it cannot. We’re planning to provide a manager (like an inner client of jobservice) that may take some responsibilities to store the stateful data in DB to make sure some data can be rebuilt from DB.

Thanks a lot

Join to automatically receive all group messages.