If a node is already powered on in the NEW state, clicking "commission node" doesn't reboot it

Bug #1386504 reported by Julian Edwards
4
This bug affects 1 person
Affects Status Importance Assigned to Milestone
MAAS
Fix Released
Wishlist
Andres Rodriguez

Bug Description

I have a node whose power is on (doesn't matter why) and it's in the NEW state.

If I click "Commission node" it says it's powering on in the event log and then later says it is powered on. However, the node is stuck at a bios prompt and actually needs rebooting to work.

I suggest we fix the templates for all the power types to cope with this. In my case, it was AMT, and it needs a "amtterm reset pxe" which would make it DTRT.

Related branches

Changed in maas:
status: New → Triaged
importance: Undecided → High
milestone: none → next
Revision history for this message
Christian Reis (kiko) wrote :

What happens to nodes that are in the NEW state that attempt to PXE boot? Do they get provided the poweroff PXELINUX image as per Blake's shutdown code?

Revision history for this message
Christian Reis (kiko) wrote :

The reason I ask is if they do, then I think the only reason you are observing this bug is because of the NUC hang bug 1376716.

There may still be an underlying issue we want to fix (ensuring that power states are always sane when we issue an action) but that may be quite minor if the exposure window is very narrow.

Revision history for this message
Christian Reis (kiko) wrote :

If we /are/ going to reboot the node, then it's probably easier and nicer to call that out to the end user, i.e. "This node is currently powered on. Shut it down first, then commission."

Revision history for this message
Julian Edwards (julian-edwards) wrote :

They do get a poweroff, yes, but that's irrelevant IMO, as is the NUC hang bug.

The power can be left on inadvertently for any number of reasons (such as someone who just installed a rack of machines and left them on after configuring BIOSes).

We *could* tell the user to shut it down but I don't think that's helpful, I'd much rather see MAAS just handle the situation and reboot it, because that's what the end user is expecting it to do.

Revision history for this message
Christian Reis (kiko) wrote :

Not if the user is doing something with the machine and mistakenly triggers the action (might be looking at the wrong node, etc) at which point you have dataloss (or worse, say BIOS upgrade, though that's a bit contrived as I'm not sure BIOS upgrades would allow AMT responses)

Changed in maas:
milestone: next → 1.7.1
Revision history for this message
Julian Edwards (julian-edwards) wrote :

If you're worried about spurious rebooting then we should put confirmation prompts in everywhere. I still think it's a little daft to allow users to just power off a deployed node without a graceful shutdown.

Changed in maas:
milestone: 1.7.1 → 1.7.2
Changed in maas:
milestone: 1.7.2 → 1.7.3
Changed in maas:
importance: High → Wishlist
milestone: 1.7.3 → 2.1.0
Changed in maas:
milestone: 2.1.0 → 2.1.1
Revision history for this message
Julian Edwards (julian-edwards) wrote :

LOL at all the release punting.

Changed in maas:
milestone: 2.1.1 → 2.1.2
Changed in maas:
milestone: 2.1.2 → 2.1.3
Changed in maas:
milestone: 2.1.3 → 2.2.0
Changed in maas:
assignee: nobody → Andres Rodriguez (andreserl)
status: Triaged → In Progress
Changed in maas:
status: In Progress → Fix Committed
Changed in maas:
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.