The OpenStack network_config.json implementation fails on Hyper-V compute nodes

Hi,
I've subscribed Xiang to this as he recently pinged me on a different string that may appear as a network device. My response to him was:

| This non-sense really needs to stop.
| We need to fix openstack to stop sending arbitrary "types" of network
| devices that mean nothing to the guest.
|
| No *new* ones should be allowed.
|
| 'vhostuser' or 'ovs' means nothing to the guest. They just see a nic.
| They can't possibly use that information in any way, so telling them is
| not helpful. The type of the device should be 'tap' or 'ethernet'.
|
| Can you submit a merge proposal upstream that does that?
|
| We can take these things in, but they're silly and quite obviously busted,
| unless you have some information that shows why they're not.

I'm willing to take this, but lets *please* work to fix the source
of the problem.

Adrian,
Can you please file a merge proposal upstream to fix this?

You're welcome to use this bug. I've made it "Also affects nova".

Fix proposed to branch: master
Review: https://review.openstack.org/400883

I've put up a request at https://review.openstack.org/400883

Hello,

as the exposing behavior for nova is like this since a few releases, it is hard to believe they will change it, due to the backwards compatibility. Basically a few stable OpenStack releases(Liberty, Mitaka, Newton, Ocata) will be probably be stuck with it :(

@Scott, thanks for your fixing! BTW, would this cloud-init version target to xenial later?

Hello Adrian, or anyone else affected,

Accepted cloud-init into xenial-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/cloud-init/0.7.8-49-g9e904bb-0ubuntu1~16.04.2 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed.Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, and change the tag from verification-needed to verification-done. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed. In either case, details of your testing will help us make a better decision.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance!

Adrian, Xiang,

Could you please verify this and mark 'verification-done' ?

At this point, this bug is blocking the release of cloud-init 0.7.8-49-g9e904bb-0ubuntu1~16.04.2 from xenial-proposed. That change contains fixes for other bugs that we need to get into -updates.

I've made requests off-bug to both Xiang Hui and to Adrian Vladu, but have not gotten a response.

Adrian has ACKed the upstream merge proposal at [1] with this fix.

While the code change does change behavior, the chance for regression is very low. See the code that was changed in context at [2]. Basically we extended the list of "physical types" to add 'hw_veb', 'hyperv', 'vhostuser'. Previously, if that condition did not match, then we would raise a ValueError exception that is not handled, leaving the system basically un-usable. Now, the strings are considered valid as "physical" and cloud-init will configure the devices as needed.

So:
  Before: cloud-init raise exception and no user-data or metadata is used... user cannot log into system.
  After: cloud-init configures networking and user-data and metadata is used.

Worst case for regression is really "still doesn't work".

--
[1] https://code.launchpad.net/~smoser/cloud-init/+git/cloud-init/+merge/311548
[2] https://git.launchpad.net/cloud-init/tree/cloudinit/sources/helpers/openstack.py#n599

Tested version 0.7.8-49-g9e904bb-0ubuntu1~16.04.2 on an OpenStack Mitaka install running Hyper-V as compute host.

VM booted successfully and cloud-init finished its run. The following output is from inside the VM after accessing it via SSH:

https://paste.ubuntu.com/23653864/

This bug was fixed in the package cloud-init - 0.7.8-49-g9e904bb-0ubuntu1~16.04.2

---------------
cloud-init (0.7.8-49-g9e904bb-0ubuntu1~16.04.2) xenial-proposed; urgency=medium

  * cherry-pick 18203bf: disk_setup: Use sectors as unit when formatting
    MBR disks with sfdisk. (LP: #1460715)
  * cherry-pick 6e92c5f: net/cmdline: Consider ip= or ip6= on command
    line not only ip= (LP: #1639930)
  * cherry-pick 8c6878a: tests: fix assumptions that expected no eth0 in
    system. (LP: #1644043)
  * cherry-pick 2d2ec70: OpenStack: extend physical types to include
    hyperv, hw_veb, vhost_user. (LP: #1642679)

 -- Scott Moser <email address hidden> Thu, 01 Dec 2016 16:57:39 -0500

The verification of the Stable Release Update for cloud-init has completed successfully and the package has now been released to -updates. Subsequently, the Ubuntu Stable Release Updates Team is being unsubscribed and will not receive messages about this bug report. In the event that you encounter a regression using the package from -updates please report a new bug using ubuntu-bug and tag the bug report regression-update so we can easily find any regressions.

This is fixed in cloud-init 0.7.9.

Hello Adrian, or anyone else affected,

Accepted cloud-init into yakkety-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/cloud-init/0.7.8-68-gca3ae67-0ubuntu1~16.10.1 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed.Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, and change the tag from verification-needed to verification-done. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed. In either case, details of your testing will help us make a better decision.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance!

Hi,
I'm going to mark this as verification-done as the original opener has not been able to do that, unfortunately. The change that went in for this fix can be seen at [1]. It is exactly the change that is in trunk, zesty, and also verified in stable release 16.04.

If it turns out that some interaction with yakkety made it *not* work, then we can re-address that.

If an sru team member wishes to disagree with my argument above, please just set it back to verification-needed, and I will attempt to get someone to do that.

Scott

--
[1] https://git.launchpad.net/cloud-init/commit/?h=ubuntu/yakkety&id=2d2ec70f06015f0624f1d0d328cc97f1fb5c29de

Because this SRU includes a large number of other bugfixes that have been verified in yakkety, we have confidence that the package is not fundamentally broken, and as you say this change has been verified on other releases, so I'm willing to accept this for the present SRU.

(But I am not releasing it on a Friday.)

Hello,

sorry for the delay, we have successfully tested a latest yakkety image(we updated via chroot the cloud-init with the one from the -proposed repo).

Thanks,
Adrian Vladu

This bug was fixed in the package cloud-init - 0.7.8-68-gca3ae67-0ubuntu1~16.10.1

---------------
cloud-init (0.7.8-68-gca3ae67-0ubuntu1~16.10.1) yakkety; urgency=medium

  * debian/cherry-pick: add utility for cherry picking commits from upstream
    into patches in debian/patches.
  * New upstream snapshot.
    - mounts: use mount -a again to accomplish mounts (LP: #1647708)
    - CloudSigma: Fix bug where datasource was not loaded in local search.
      (LP: #1648380)
    - when adding a user, strip whitespace from group list
      [Lars Kellogg-Stedman] (LP: #1354694)
    - fix decoding of utf-8 chars in yaml test
    - Replace usage of sys_netdev_info with read_sys_net (LP: #1625766)
    - fix problems found in python2.6 test.
    - OpenStack: extend physical types to include hyperv, hw_veb, vhost_user.
      (LP: #1642679)
    - tests: fix assumptions that expected no eth0 in system. (LP: #1644043)
    - net/cmdline: Consider ip= or ip6= on command line not only ip=
      (LP: #1639930)
    - Just use file logging by default [Joshua Harlow] (LP: #1643990)
    - Improve formatting for ProcessExecutionError [Wesley Wiedenmeier]
    - flake8: fix trailing white space
    - Doc: various documentation fixes [Sean Bright]
    - cloudinit/config/cc_rh_subscription.py: Remove repos before adding
      [Brent Baude]
    - packages/redhat: fix rpm spec file.
    - main: set TZ in environment if not already set. [Ryan Harper]
    - disk_setup: Use sectors as unit when formatting MBR disks with sfdisk.
      [Daniel Watkins] (LP: #1460715)

 -- Scott Moser <email address hidden> Mon, 19 Dec 2016 15:07:12 -0500

this is definitely *not* fix-released in nova.
We see more bugs like: bug 1674946

Reviewed: https://review.openstack.org/400883
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=f559be35a03f5801f527355895a97c89cdc3c336
Submitter: Jenkins
Branch: master

commit f559be35a03f5801f527355895a97c89cdc3c336
Author: Scott Moser <email address hidden>
Date: Fri Mar 31 17:01:33 2017 -0400

    Limit exposure of network device types to the guest.

    Previously, the 'type' of the hypervisor network device, was exposed to
    the guest directly. That does not make sense, as
    a.) this leaks needless information into the guest
    b.) the guest cannot be reasonably expected to make decisions
        based on a type of link that is present underneath the
        virtual device that is presented to the guest.
    c.) guests then are forced to either continuously track these types
        or to assume that unknown type is "phy".

    This limits the exposure of types to a specific list. Any other
    type will be shown to the guest as 'phy'.

    Change-Id: Iea458fba29596cd2773d8d3565451af60b02bcca
    Closes-Bug: #1642679

This issue was fixed in the openstack/nova 16.0.0.0b1 development milestone.

I still hit this issue on latest xenial cloudimg of April 25th. This is the error I saw when trying to run an Ubuntu 16.04 guest OS on a contrail based cloud: http://paste.openstack.org/show/608110/

Fix proposed to branch: stable/ocata
Review: https://review.openstack.org/476195

Guys we still hitting the same bug, this started to be a huge issue for us.
http://paste.openstack.org/show/615013/
http://paste.openstack.org/show/615127/
Each time we run vm with ubuntu 16.04 we randomly see this bug in vm boot-log.
OpenStack: Mitaka
Juniper Contrail: 3.2.8
Please provide fix for that.

Reviewed: https://review.openstack.org/476195
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=cec7ecdc93c3b9ba401edf3cf84088b580247cb8
Submitter: Jenkins
Branch: stable/ocata

commit cec7ecdc93c3b9ba401edf3cf84088b580247cb8
Author: Scott Moser <email address hidden>
Date: Fri Mar 31 17:01:33 2017 -0400

    Limit exposure of network device types to the guest.

    Previously, the 'type' of the hypervisor network device, was exposed to
    the guest directly. That does not make sense, as
    a.) this leaks needless information into the guest
    b.) the guest cannot be reasonably expected to make decisions
        based on a type of link that is present underneath the
        virtual device that is presented to the guest.
    c.) guests then are forced to either continuously track these types
        or to assume that unknown type is "phy".

    This limits the exposure of types to a specific list. Any other
    type will be shown to the guest as 'phy'.

    Change-Id: Iea458fba29596cd2773d8d3565451af60b02bcca
    Closes-Bug: #1642679
    (cherry picked from commit f559be35a03f5801f527355895a97c89cdc3c336)

This issue was fixed in the openstack/nova 15.0.7 release.

Tracked in Github Issues as https://github.com/canonical/cloud-init/issues/2761