Project

Profile

Help

Task #6624

closed

Document how Pulp handles the ServerDisconnectedError and why

Added by ttereshc over 4 years ago. Updated almost 3 years ago.

Status:
CLOSED - DUPLICATE
Priority:
Normal
Assignee:
-
Category:
-
Sprint/Milestone:
-
Start date:
Due date:
% Done:

0%

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

Description

Ticket moved to GitHub: "pulp/pulpcore/1888":https://github.com/pulp/pulpcore/issues/1888


A new FAQ page

Create a new FAQ page and have this be the first FAQ. This should be a top-level page on the left-side of the navigation.

Behavior to document

At open floor today we touched on how Pulp 3 does not have retry logic in these cases:

  • the server hanging up the TCP connection
  • network errors which cause TCP hangups
  • http errors other than [429, 502, 503, 504]

Reasoning why

Two issues would occur if Pulp had a broad retry behavior such as in the cases above:

  1. If the server says stop by hanging up Pulp should stop. Similarly if the network delivers TCP segments so slowly TCP times our.

  2. If pulp did continue an actual network error would become a very slow failure, tons of retries for each content unit slowly failing one by one over hours

What can users who want retry do?

Users can restart their sync tasks. Pulp saves all the content it did receive on the timed-out sync, so the next time it syncs it doesn't need to redownload what it already got. Therefore when users do retry they effectively resume whenever the network and server is ready for that.


Related issues

Related to RPM Support - Issue #6589: Issues synching large Redhat reposCLOSED - DUPLICATEActions

Also available in: Atom PDF