Project

Profile

Help

Story #7201

As a user I can set http connection timeout and read timeout

Added by bmbouter 3 months ago. Updated 3 months ago.

Status:
NEW
Priority:
Normal
Assignee:
-
Category:
-
Start date:
Due date:
% Done:

0%

Estimated time:
Platform Release:
Groomed:
No
Sprint Candidate:
No
Tags:
Katello
Sprint:
Quarter:

Description

Problem

In Pulp2 it is possible to set the hp connection timeout and read timeouts https://docs.pulpproject.org/en/2.21/user-guide/server.html?highlight=connect_timeout. This recently helped me resolve an issue an with a few repos in Pulp that were failing to sync. In Pulp3 documentation setting these timeouts is discussed in https://docs.pulpproject.org/plugins/api-reference/download.html?highlight=timeout, however it is only if you are a plugin in developer.

As a user I need to be able to set this for my environment.

Solution

Add these timeouts to BaseRemote so they would automatically be available to all plugins.

  • total_timeout - This is a float field, measured in second. This would drive the ClientTimeout.total field in aiohttp. Pulp default is 5 min. None or 0 disables a particular timeout check.
  • connect_timeout - This is a float field, measured in second. This would drive the ClientTimeout.connect field in aiohttp. Pulp default is None. None or 0 disables a particular timeout check.
  • sock_connect_timeout - This is a float field, measured in second. This would drive the ClientTimeout.sock_connect field in aiohttp. Pulp default is None. None or 0 disables a particular timeout check.
  • sock_read_timeout - This is a float field, measured in second. This would drive the ClientTimeout.sock_read field in aiohttp. Pulp default is None. None or 0 disables a particular timeout check.

History

#1 Updated by bmbouter 3 months ago

  • Description updated (diff)

#2 Updated by bmbouter 3 months ago

  • Sprint set to Sprint 77

I recreated this issue, it was accidentally deleted.

@mped, here's the text you had posted:

I think this would also help with issues I was seeing in: https://pulp.plan.io/issues/6589. The reason for this being, during this issue in one of the envrionments I was coming from behind a proxy, and this is where this issue looked a lot worse.

I have since tested to see what happened with a new Pulp2 server (I know they are not the same but it was just to see if the environment was causing issues). Also ended up with quite a few issues with Pulp2, but setting the connect_timeout and read_timeout resolved the issues, as I was seeing a lot of problems with packages failing to being downloaded/retried and then the task cancelling itself.

#3 Updated by rchan 3 months ago

  • Sprint changed from Sprint 77 to Sprint 78

#4 Updated by daviddavis 3 months ago

  • Tags Katello added

#5 Updated by daviddavis 3 months ago

Removing per sprint planning.

#6 Updated by rchan 3 months ago

  • Sprint deleted (Sprint 78)

Please register to edit this issue

Also available in: Atom PDF