Stored vs Instance Reported Address Comparison Causes Repeated Address Updates

[Joseph Phillips]

When the instance poller checks the provider-sourced addresses and compares them with what we have stored in state they are always unequal, causing addresses to be repeatedly updated (at least for MAAS).

This is because the stringified provider address includes a space ID, but what we compare it to does not.

For example:

"local-cloud:172.16.99.3@space-default(id:1)"
versus
"local-cloud:172.16.99.3@space-default"

Fixing this will reduce DB load, as it is running every 5 seconds for each machine.

[John A Meinel (jameinel)]

Thanks for tracking this down, Joe. We've had issues similar to this in the
past but this is a different flavor of it. Certainly worth fixing.

John
=:->

On Tue, May 14, 2019, 11:11 Joseph Phillips <email address hidden>
wrote:

> Public bug reported:
>
> When the instance poller checks the provider-sourced addresses and
> compares them with what we have stored in state they are always unequal,
> causing addresses to be repeatedly updated (at least for MAAS).
>
> This is because the stringified provider address includes a space ID,
> but what we compare it to does not.
>
> For example:
>
> "local-cloud:172.16.99.3@space-default(id:1)"
> versus
> "local-cloud:172.16.99.3@space-default"
>
> Fixing this will reduce DB load, as it is running every 5 seconds for
> each machine.
>
> ** Affects: juju
> Importance: High
> Assignee: Joseph Phillips (manadart)
> Status: In Progress
>
> ** Affects: juju/2.5
> Importance: High
> Assignee: Joseph Phillips (manadart)
> Status: In Progress
>
> ** Affects: juju/2.6
> Importance: High
> Assignee: Joseph Phillips (manadart)
> Status: In Progress
>
> ** Changed in: juju
> Status: New => In Progress
>
> ** Changed in: juju
> Importance: Undecided => High
>
> ** Changed in: juju
> Assignee: (unassigned) => Joseph Phillips (manadart)
>
> ** Changed in: juju
> Milestone: None => 2.7-beta1
>
> ** Also affects: juju/2.6
> Importance: Undecided
> Status: New
>
> ** Also affects: juju/2.5
> Importance: Undecided
> Status: New
>
> ** Changed in: juju/2.5
> Status: New => In Progress
>
> ** Changed in: juju/2.6
> Status: New => In Progress
>
> ** Changed in: juju/2.5
> Importance: Undecided => High
>
> ** Changed in: juju/2.6
> Importance: Undecided => High
>
> ** Changed in: juju/2.5
> Assignee: (unassigned) => Joseph Phillips (manadart)
>
> ** Changed in: juju/2.6
> Assignee: (unassigned) => Joseph Phillips (manadart)
>
> ** Changed in: juju/2.5
> Milestone: None => 2.5.8
>
> ** Changed in: juju/2.6
> Milestone: None => 2.6.2
>
> --
> You received this bug notification because you are subscribed to juju.
> Matching subscriptions: juju bugs
> https://bugs.launchpad.net/bugs/1828972
>
> Title:
> Stored vs Instance Reported Address Comparison Causes Repeated Address
> Updates
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/juju/+bug/1828972/+subscriptions
>

[Joseph Phillips (manadart)]

2.5: https://github.com/juju/juju/pull/10182

[Joseph Phillips (manadart)]

2.6: https://github.com/juju/juju/pull/10188

[Joseph Phillips (manadart)]

2.6 above was aborted due to branch churn at the time.
New patch is: https://github.com/juju/juju/pull/10192

[John A Meinel (jameinel)]

Didn't this get brought in via https://github.com/juju/juju/pull/10190 ?

On Wed, May 15, 2019 at 11:35 AM Joseph Phillips <email address hidden>
wrote:

> 2.6 above was aborted due to branch churn at the time.
> New patch is: https://github.com/juju/juju/pull/10192
>
> ** Changed in: juju/2.6
> Status: In Progress => Fix Committed
>
> --
> You received this bug notification because you are subscribed to juju.
> Matching subscriptions: juju bugs
> https://bugs.launchpad.net/bugs/1828972
>
> Title:
> Stored vs Instance Reported Address Comparison Causes Repeated Address
> Updates
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/juju/+bug/1828972/+subscriptions
>

[Joseph Phillips (manadart)]

No; I closed 10188 at my EoD when 2.6 was hard reset. The merge was re-submitted as 10192 the next day when the dust had cleared.

[Joseph Phillips (manadart)]

Develop: https://github.com/juju/juju/pull/10194