Can't deploy to Dell R430 server

I’m trying to deploy a Juju controller to a Dell R430 server. MAAS sees the server, and it’s able to power it on/off over IPMI. The MAAS web interface shows it going through an initial boot (ephemeral), but a later reboot sticks for a long time, then seems to time out. On the console for that server, the second PXE boot stops at a “boot:” prompt.

Any ideas?

I have one idea :upside_down_face:
EFI or Legacy boot?

Hi, @rainwadj,

We need some more information.

Can you please identify your version and build

We need to know the version and build (and packaging format) that you’re running. Here’s how to get that info:

If you’re using a snap

If you’re using a snap, execute snap listmaas at the command line, which will return some lines like this:

Name  Version                       Rev    Tracking     Publisher   Notes
maas  3.0.0~beta2-9796-g.2182ab55f  13292  latest/edge  canonical✓  -

Please add the output of that command to a comment on this post.

If you’re using a Debian package

If you’re using a deb, execute apt list maas at the command line, and enter whatever it returns into a comment to this post, as in the snap example above.

Using CLI, UI, or API?

Next, you’ll need to specify which interface you’re using, and generally what command(s) you were attempting.

Explain what happens

Being as concise and specific as you can, please explain a little more about what seemed to go wrong. You mentioned that it hangs at the second PXE boot prompt. Is there anything to indicate that it can’t find an image for that second boot, for example?

Try and explain how to reproduce your issue

In addition to the explanation above, you should create a step-by-step list of what you did to reproduce the problem. This might be a little complicated with Juju, but anything you can tell us may help.

Any relevant screenshots?

Can you get a console on the R430 to watch the boot process, or lack thereof? If so, screenshots might save a lot of typing. You can just upload them into your comment.

Locate and capture logfiles

I dunno if any logfiles will help with this, but you might check these:

  1. maas.log
  2. regiond.log
  3. rackd.log
  4. the rsyslog file of the affected machine(s), if it exists.

On snap, they’re in:

  • /var/snap/maas/common/log/maas.log
  • /var/snap/maas/common/log/regiond.log
  • /var/snap/maas/common/log/rackd.log
  • /var/snap/maas/common/log/rsyslog/$MACHINE_NAME/$RELEVANT_DATE/messages

If you’re using Debian packages, they’re somewhere else:

  • /var/log/maas/maas.log
  • /var/log/maas/regiond.log
  • /var/log/maas/rackd.log
  • /var/log/maas/rsyslog/$MACHINE_NAME/$RELEVANT_DATE/messages

Ideally, you’d want to put these in a pastebin and link the pastebin location in your comment.

Thanks for this. We’ll very likely need this additional info to sort this.

1 Like

Hello, @billwear

I’m trying a similar problem with Asus X-99-WS/IPMI physical machines. Maas doesn’t recognize the machines and I’m trying manually. In this phase, Maas activates Node Power ON but after 30 minutes the commissioning is frustrated, due to failure to access the BMC. Through the URL I access ASMB8-iKVM and also via CLI impitools, no difficulty. The administrator passwords are then correct.

Name Version Rev Tracking Publisher Notes
Maas 3.2.6-12016-g.19812b4da 23947 3.2/stable canonical✓ -

I am mainly using UI and some commands via CLI, based on the tutorial of this link: https://blog.naydenov.net/2015/11/deploying-openstack-on-maas-1-9-with-juju-network-setup/

The prints and log files are at this link: https://drive.pen6.link/s/8A5Xpb45qJQkDs9

Thank you very much if you can help me.

Hello @iaginde1

I think it can be both one and the other, they just need to be the same, both in the bios and in the maas. That is, if it is legacy in the bios, it should be legacy in the maas, and be efi in the bios, so it should also be in the maas.

But, I can be corrected, in case someone has more information.

Greetings.

Hi @penacleiton,

Have you tried booting up nodes in PXE without configuring IPMI manually in maas?

It looks like you add the devices yourself, try let them boot up into PXE and let MAAS do it’s magic, after that the nodes should appear as a new machine under “machines”.

1 Like

Hi @penacleiton

Yes, if I understand you correctly, you’re trying to initialize physical machine that already have an OS’s installed?

Maas should be the one deploying OS’s for you, so no need to install it before hand.
In my testing environment I almost got the same setup as you, 6 hosts plus 1 host with Maas controller+rack (3.2/stable).

The controller is configured with 2 subnets, one for PXE and one for external usage. The PXE subnet is configured with DHCP inside maas controller. The external subnet is handled by the router.

All 6 hosts have been configured with static IP for BMC/IPMI, PXE as the first boot priority and UEFI only.

Commissioning new machines: I then start the hosts one by one to let them PXE boot. If the network configuration is right, maas should see the host requesting and send the boot files. It then loads the OS inside the RAM to initialize it and when it’s done it should appear as a “new” machine under “machines”. You can then commission “new” machines and after that they should be “ready” (for you to ex. deploy an OS on it).

tag: @erik-lonroth

2 Likes

Thank you for your reply, @Marcus
In fact, you understood perfectly, and I was the one who had not understood your comment about PXE. I apologize for that, but I see that now you have made everything much more transparent when comparing it with your test experiment.

I need to understand this better. I think it’s getting much clearer now.

Since I previously installed Ubuntu 22.04 both on the controlling machine and on all 6 physical machines, do you recommend that I uninstall Ubuntu on the 6 nodes and connect to the LAN Management port of the motherboard Asus X-99-WS, without any operating system installed?

Once this is done, will the MAAS that is installed on the controlling machine recognize and install the operating system automatically, on all physical machines? Ou nem preciso desinstalar a versão do Ubuntu que está nas máquinas?

Is that it?

Thank you, once again.

2 Likes

Good day @penacleiton!

do you recommend that I uninstall Ubuntu on the 6 nodes?

No need for uninstalling, you will be reinstalling them later when you got them assigned to maas.

connect to the LAN Management port of the motherboard Asus X-99-WS?

Make sure that the nodes are connected to same network as MAAS PXE network.

When maas is initializing a node, it creates a “maas” user on the bmc/ipmi so it can control it (eg. power, don’t know if more.)

Once this is done, will the MAAS that is installed on the controlling machine recognize and install the operating system automatically, on all physical machines?

MAAS controller will be able to control power on/off, deploying OS’s, pre storage partioning and networking.

This is an example I used when learning https://youtu.be/b_bdVfG47G4 and of course the documentation https://maas.io/docs

Please reply how it goes for you.

2 Likes

@penacleiton - I’m glad we can help you. @marcus is working with our team and we are currently setting up multiple environments like the one you are experimenting with.

1 Like

Thank you @erik-lonroth. Happy New Year :pray:

1 Like

Thank you @Marcus. Happy New Year! :pray:

2 Likes

Hi @Marcus

I made great progress thanks to your tips. I sent a message explaining the details. I look forward to your reply.

Greetings.

@rainwadj, did you ever get this resolved? see the chain of commments for more information and suggestions, but lmk if this is still an issue for you.