Details
-
Type:
Task
-
Status: Open
-
Priority:
Minor
-
Resolution: Unresolved
-
Component/s: etc
-
Labels:None
Description
There appears to be some delay before mirrors get updated data files, and this timing appears to be different from one mirror to another.
Coupled with the random forwarding from mirror brain, the user can see inconsistent data — imagine there's mirror X serving updates for 1.405 and another mirror Y that's still serving Y. A Jenkins instance out there can see X, then he can refresh the update center metadata and now see the old data from Y.
I wonder if mirror brain can be configured to forward based on consistent hash from user's IP address or something, which would consistently forward requests from the same user to the same host (but without losing the redundancy in case of a host failure.)
Activity
| Field | Original Value | New Value |
|---|---|---|
| Project | Jenkins [ 10172 ] | Infrastructure [ 10301 ] |
| Key | JENKINS-9245 | INFRA-160 |
| Workflow | JNJira [ 139406 ] | jira [ 158522 ] |
| Component/s | etc [ 18774 ] | |
| Component/s | infrastructure [ 15687 ] |