Skip to content
This repository was archived by the owner on Feb 24, 2020. It is now read-only.

stage1: update to CoreOS 1353.0.0 and systemd v233#3619

Closed
dm0- wants to merge 1 commit intorkt:masterfrom
dm0-:v233
Closed

stage1: update to CoreOS 1353.0.0 and systemd v233#3619
dm0- wants to merge 1 commit intorkt:masterfrom
dm0-:v233

Conversation

@dm0-
Copy link
Contributor

@dm0- dm0- commented Mar 15, 2017

Here is the change I used to try systemd v233 with rkt. Sort of. I actually used a local PXE image since the 1353.0.0 release isn't until tomorrow.

This partially supersedes #3535.

@lucab
Copy link
Member

lucab commented Mar 16, 2017

(to be retriggered once PXE image is up)

@lucab lucab mentioned this pull request Mar 16, 2017
Closed
@dm0-
Copy link
Contributor Author

dm0- commented Mar 16, 2017

The PXE images should be up for both architectures.

@euank
Copy link
Member

euank commented Apr 3, 2017

The test failure:

19:50:23 --- FAIL: TestVolumeSysfs (9.63s)
19:50:23 	rkt_volume_nspawn_test.go:76: Running test #0: Without any mount
19:50:23 	rkt_volume_nspawn_test.go:76: Running test #1: With /sys as an empty volume
19:50:23 	rkt_volume_nspawn_test.go:76: Running test #2: With /sys bind-mounted from the host
19:50:23 	rkt_volume_nspawn_test.go:76: Running test #3: With /sys/class bind-mounted from the host
19:50:23 	rkt_tests.go:410: didn't receive expected output "/sys/class: mode: Dcrw-rw-rw-": image: using image from local store for image name coreos.com/rkt/stage1-coreos:1.25.0+git4fbe518

...

19:50:23 		[�[0;32m  OK  �[0m] Started rkt supervisor-ready signaling.
19:50:23 		[ 2298.773944] rkt-inspect[5]: /sys/class: mode: drwxr-xr-x
19:50:23 		[ 2298.774211] rkt-inspect[5]: /sys/class: user: 0
19:50:23 		[ 2298.774368] rkt-inspect[5]: /sys/class: group: 0
19:50:23 		Container rkt-6fc0ed05-750e-492c-8291-2e1c8274db48 exited successfully.

@dm0-
Copy link
Contributor Author

dm0- commented Apr 7, 2017

@iaguis Are you familiar with the meaning of that mode? I don't see why /sys/class would ever be reported to be a character device.

@euank
Copy link
Member

euank commented Apr 7, 2017

@dm0- it looks like it's actually trying to bindmount /dev/null over it is why.

https://github.com/rkt/rkt/blame/cf0aa2e4c6062e76c6a18a8af99641e84aac9ecc/tests/rkt_volume_nspawn_test.go#L61-L62

Seems like nspawn or systemd might be clobbering the user's chosen mount there for whatever reason.

Maybe the masked paths thing, which we had a hidden knob to opt out of iirc, is back.

@dm0- dm0- closed this Oct 4, 2017
@dm0- dm0- deleted the v233 branch October 4, 2017 20:23
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Reviewers

No reviews

Assignees

No one assigned

Labels

component/stage1 needs/tests

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@dm0- @lucab @euank