Public address of unit changes take a while to update
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Canonical Juju |
Fix Released
|
Medium
|
Achilleas Anagnostopoulos | ||
2.6 |
Won't Fix
|
Medium
|
Achilleas Anagnostopoulos |
Bug Description
Hi,
For units deployed in public clouds (AWS, Azure, etc.), the public address assigned is only for the duration that the unit is up and running. People often stop instances and start them back up, for example to address planned maintenance work by AWS or to resize to different flavors. Unfortunately, that then assigns new or different public addresses to these units which juju take ~10 mins or so to notice.
Any chance this could be noticed quicker? unit down + lost agent, then back up could perhaps inform the controller of the unit's new public address.
Or failing that, some command to force manual refresh of the cached public address?
This is two AWS environments running Juju 2.6.10.
To reproduce, spin up an instance or two. Login to the public cloud dashboard/portal, stop instance, wait, start. Instance should boot with different/new IPs. 'juju status' should show the old still until about ~10mins.
Changed in juju: | |
status: | New → Triaged |
importance: | Undecided → Medium |
milestone: | none → 2.8-beta1 |
Changed in juju: | |
assignee: | nobody → Canonical Juju QA Bot (juju-qa-bot) |
Changed in juju: | |
assignee: | Canonical Juju QA Bot (juju-qa-bot) → nobody |
assignee: | nobody → Achilleas Anagnostopoulos (achilleasa) |
description: | updated |
description: | updated |
Changed in juju: | |
milestone: | 2.7-rc4 → none |
Changed in juju: | |
status: | Fix Committed → Fix Released |
Issue affects both 2.6.x and 2.7-rcX. Working on a potential fix.