Switch default instance type from m1.small to t2.small/m3.medium for EC2 provider

Bug #1373516 reported by Robert C Jennings
14
This bug affects 2 people
Affects Status Importance Assigned to Milestone
juju-core
Fix Released
High
Cheryl Jennings
1.24
Fix Released
Critical
Cheryl Jennings
1.25
Fix Released
High
Cheryl Jennings

Bug Description

m1.small as the default instance size relies on the previous generation instance type [1]. Amazon is encouraging moving to current generation instance types. t2.small would be comparable to m1.small at half the price or m3.medium would provide consistent performance for a bit more. Either way, my testing is failing because capacity for m1.small is being exhausted.

$ juju charm test -e amazon-charmtest
juju-test INFO : Starting test run on amazon-charmtest using Juju 1.20.7
Launching instance
ERROR bootstrap failed: cannot start bootstrap instance: cannot run instances: We currently do not have sufficient m1.small capacity in the Availability Zone you requested (us-east-1a). Our system will be working on provisioning additional capacity. You can currently get m1.small capacity by not specifying an Availability Zone in your request or choosing us-east-1d, us-east-1b. (InsufficientInstanceCapacity)
Bootstrap failed, destroying environment
ERROR cannot start bootstrap instance: cannot run instances: We currently do not have sufficient m1.small capacity in the Availability Zone you requested (us-east-1a). Our system will be working on provisioning additional capacity. You can currently get m1.small capacity by not specifying an Availability Zone in your request or choosing us-east-1d, us-east-1b. (InsufficientInstanceCapacity)
juju-test.conductor WARNING : Could not bootstrap amazon-charmtest, got Bootstrap returned with an exit > 0. Skipping
Launching instance
ERROR bootstrap failed: cannot start bootstrap instance: cannot run instances: We currently do not have sufficient m1.small capacity in the Availability Zone you requested (us-east-1a). Our system will be working on provisioning additional capacity. You can currently get m1.small capacity by not specifying an Availability Zone in your request or choosing us-east-1d, us-east-1b. (InsufficientInstanceCapacity)
Bootstrap failed, destroying environment
ERROR cannot start bootstrap instance: cannot run instances: We currently do not have sufficient m1.small capacity in the Availability Zone you requested (us-east-1a). Our system will be working on provisioning additional capacity. You can currently get m1.small capacity by not specifying an Availability Zone in your request or choosing us-east-1d, us-east-1b. (InsufficientInstanceCapacity)
juju-test.conductor WARNING : Could not bootstrap amazon-charmtest, got Bootstrap returned with an exit > 0. Skipping
juju-test INFO : Results: 0 passed, 0 failed, 2 errored

[1] https://aws.amazon.com/blogs/aws/aws-update-new-m3-features-reduced-ebs-prices-reduced-s3-prices/

Tags: ec2-provider
Robert C Jennings (rcj)
summary: - Switch default instance type from m1 to m3 for EC2 provider
+ Switch default instance type from m1.small to t2.small/m3.medium for EC2
+ provider
Curtis Hovey (sinzui)
tags: added: ec2-provider
Changed in juju-core:
status: New → Triaged
importance: Undecided → Medium
Revision history for this message
Kapil Thangavelu (hazmat) wrote :

sigh.. agreed this should be t2.small/m3.medium but the issue on zone avail is separate and not related (and also filed separately).

Revision history for this message
Kapil Thangavelu (hazmat) wrote :

actually take that bake a bit, the larger issue with t2.small is that its a hypervisor penalty instance which is poorly suited for a state server.

Revision history for this message
Jorge Castro (jorge) wrote :

Just ran into this again, we can prioritize this?

Revision history for this message
Jorge Castro (jorge) wrote :

Also AWS has recommended m3's for all new workloads: https://aws.amazon.com/blogs/aws/ec2-update-previous-generation-instances/

They've also dropped all the m1's from their web console.

Revision history for this message
Ben Howard (darkmuggle-deactivatedaccount) wrote :

Raising this to critical. IMHO, this is going to be a substantial problem for our users as they will hit capacity constraints as the availability of m1.x family becomes limited. Further, users are going to hit issues with performance since the newer instance types are more performant and cheaper.

Changed in juju-core:
importance: Medium → Critical
Ian Booth (wallyworld)
Changed in juju-core:
milestone: none → 1.25-beta1
milestone: 1.25-beta1 → 1.26-alpha1
Revision history for this message
Nate Finch (natefinch) wrote :

Working on this now.

Changed in juju-core:
assignee: nobody → Nate Finch (natefinch)
Ian Booth (wallyworld)
no longer affects: juju-core/1.26
Revision history for this message
Nate Finch (natefinch) wrote :

For reference, I'm going to set the default to m3.medium... if you don't like that choice, speak now or forever hold your peace.

Curtis Hovey (sinzui)
Changed in juju-core:
importance: Critical → High
Changed in juju-core:
assignee: Nate Finch (natefinch) → Katherine Cox-Buday (cox-katherine-e)
Revision history for this message
Katherine Cox-Buday (cox-katherine-e) wrote :

1.25 review request is here: http://reviews.vapour.ws/r/2720/

Changed in juju-core:
status: Triaged → In Progress
Revision history for this message
Katherine Cox-Buday (cox-katherine-e) wrote :

The fix is causing some unexpected test failures. More investigation is needed, so we're pushing to beta2 for 1.25.

Changed in juju-core:
assignee: Katherine Cox-Buday (cox-katherine-e) → Cheryl Jennings (cherylj)
Changed in juju-core:
status: In Progress → Fix Committed
Curtis Hovey (sinzui)
Changed in juju-core:
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.