autoinstall (re-)using 'autoinstall-user-data' fails on impish

Bug #1946241 reported by Frank Heimes
10
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Ubuntu on IBM z Systems
Fix Released
High
Canonical Foundations Team
subiquity
Fix Released
Undecided
Unassigned

Bug Description

I tried autoinstall installations on impish (final beta) on s390x LPARs (and VLAN connectivity),
using the auto-generated 'autoinstall-user-data' from a previous interactive impish installation,
and it fails with the following error message (might be unrelated to s390x):

"- zFCP LUN 0.0.e100:0x50050763061b16b6:0x4026400300000000"
"An error occurred. Press enter to start a shell"
"generating crash report"
"report saved to /var/crash/1633527110.880342960.ui.crash"
"Traceback (most recent call last):"
"File ""/snap/subiquity/2652/lib/python3.6/site-packages/subiquity/client/client.py"", line 369, in run"
super().run()
"File ""/snap/subiquity/2652/lib/python3.6/site-packages/subiquitycore/tui.py"", line 374, in run"
super().run()
"File ""/snap/subiquity/2652/lib/python3.6/site-packages/subiquitycore/core.py"", line 128, in run"
"raise exc"
"File ""/snap/subiquity/2652/lib/python3.6/site-packages/subiquity/client/client.py"", line 240, in noninteractive_watch_app_state"
"await run_in_thread(input)"
"File ""/snap/subiquity/2652/lib/python3.6/site-packages/subiquitycore/async_helpers.py"", line 46, in run_in_thread"
"return await loop.run_in_executor(None, func, *args)"
"File ""/snap/subiquity/2652/usr/lib/python3.6/concurrent/futures/thread.py"", line 56, in run"
"result = self.fn(*self.args, **self.kwargs)"
"EOFError: EOF when reading a line"

I tried two different disk types (FCP and DASD), but things fail either way.

I've attached the console log (that is unfortunately everything that I can get, since no logon is possible).

I'll try on z/VM (and w/o VLAN) as well, trying to narrow down things.

Please notice that I was already able to perform autoinstalls with impish,
so it looks like a regression with a recent change?!

An older user-data (that was know to work in the past) still works.

The difference that is causing the problem is a 'kernel' top level key that's generated.
If removing (or commenting it out), things work again.

[With that I'm pretty sure it's not s390x specific.]

Revision history for this message
Frank Heimes (fheimes) wrote :
Revision history for this message
Frank Heimes (fheimes) wrote :
Revision history for this message
Frank Heimes (fheimes) wrote :

Interesting, so autoinstall worked in my z/VM environment.

Two main things are different:
1) normal nw connectivity (no VLAN)
2) I used n older user-data file (on the previous LPAR tests that I mentioned about I reused a newly created 'autoinstall-user-data' from a prior manual impish installation)

Next thing I'm going to try is to take the z/VM user-data file, change it so that it fits to the LPAR and try again (and figure out any differences)...

Revision history for this message
Frank Heimes (fheimes) wrote :

So it boils down to a problem with the newly auto-generated 'autoinstall-user-data' file (generated by an interactive impish installation), that obviously looks and behaves a bit different.
Trying to find the delta bit that is causing the issue,
and updating title to reflect this outcome ...

summary: - autoinstall fails on impish
+ autoinstall (re-)using 'autoinstall-user-data' fails on impish
description: updated
Changed in ubuntu-z-systems:
importance: Undecided → Medium
assignee: nobody → Canonical Foundations Team (canonical-foundations)
Revision history for this message
Frank Heimes (fheimes) wrote :

After doing a (kind of a) bi-sect of the config (user-data), I am sure that the suspect that is causing the issue is this 'kernel' top level key and config:
  kernel:
    kernel: {package: linux-generic}
If I comment out (or just delete) these lines, I no longer run into the install problem.
(I of course have to enable the ccw devices in an early-command section manually, but that is totally unrelated to this.)

This 'kernel' key is also one of the options that I don't have in my 'known as good' user-data example.

And I even cannot find anything about a 'kernel' top level key in the autoinstall reference:
https://ubuntu.com/server/docs/install/autoinstall-reference

Well, it should probably not have been auto-generated at all?!

description: updated
Revision history for this message
Michael Hudson-Doyle (mwhudson) wrote :

This should be fixed in the latest impish dailies as it happens. Which serial did you use for the install?

Revision history for this message
Frank Heimes (fheimes) wrote :

I used the image that was published as final beta:
ubuntu-21.10-beta-live-server-s390x.iso (Sep 21)
Haven't updated to any potential dailies that came after this.
May do today ...

tags: added: installer
removed: install
Revision history for this message
Frank Heimes (fheimes) wrote (last edit ):

I've just updated the install server to the daily 'current':
https://cdimage.ubuntu.com/ubuntu-server/daily-live/current/impish-live-server-s390x.iso
(which is from Oct 4th, so still some days old)
and got a notification about updating the installer from 21.08.2 to 21.10.1,
confirmed the update and it failed.
Connection dropped and I cannot login again:
https://pastebin.canonical.com/p/YXb3wrqs8g/

Trying now to install based on this image w/o updating the installer ...

Installation completed fine without updating.

The auto-generated autoinstall-user-data contains the kernel key and section.

Trying to autoinstall using a user-data file based on the above autoinstall-user-data (with having the kernel key still in) still fails.

Please see below the 'impish_install_update_oct_4th.txt' for the subiquity update error.
and 'impish_autoinstall_oct_4th.txt' for the re-try of the autoinstall based on daily (current) - both with Oct 4th.

(btw. I also tried the image from pending, Oct 6th, but an interactive installations fails with that at all, probably the reason why it didn't came out of pending state, yet)

Revision history for this message
Frank Heimes (fheimes) wrote :
Revision history for this message
Frank Heimes (fheimes) wrote :
Revision history for this message
Ubuntu QA Website (ubuntuqa) wrote :

This bug has been reported on the Ubuntu ISO testing tracker.

A list of all reports related to this bug can be found here:
http://iso.qa.ubuntu.com/qatracker/reports/bugs/1946241

tags: added: iso-testing
Dan Bungert (dbungert)
tags: added: fr-1790
Revision history for this message
Michael Hudson-Doyle (mwhudson) wrote :

I think ssh should work if you update again (and on the next ISOs, probably not the current one)

Revision history for this message
Frank Heimes (fheimes) wrote :

I still get the "ModuleNotFoundError: No module named 'subiquity'" with the impish daily (Oct 7th)
https://pastebin.ubuntu.com/p/qBRNQpz857/
I guess I have to wait for the next daily to get the fix ....

Frank Heimes (fheimes)
Changed in ubuntu-z-systems:
importance: Medium → High
Revision history for this message
Frank Heimes (fheimes) wrote :

So 1st of all I can confirm that the "ModuleNotFoundError: No module named 'subiquity'" is solved with the daily from Oct 8th (taken from pending) and I was able to complete installations usinf ssh from remote again.

A kernel key entry is still generated in /var/log/installer/autoinstall-user-data
but it's now:
  kernel:
    package: linux-generic
was formerly:
  kernel:
    kernel: {package: linux-generic}
And with this change the auto install 'round-trip' (re-using /var/log/installer/autoinstall-user-data) works again.

Many thx ! Closing this bug as Fix Released.

Changed in subiquity:
status: New → Fix Released
Changed in ubuntu-z-systems:
status: New → 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.