Red Hat Bugzilla – Bug 1455455
[RFE] PXE less provisioning - Add delay to discovery image boot for slow DHCP networks
Last modified: 2018-02-21 07:39:11 EST
Description of problem: Booting of PXE less discovery image is failing with network initialization error, because of slow DHCP network. How reproducible: On customers side. Steps to Reproduce: 1. create a discovery image 2. boot the server 3. wait for error Actual results: Boot fails with network initialization error. Expected results: Discovery image will boot. Additional info: Adding sleep(15) at the top of the configure_network function workarounds the issue. According to lzap we should include delay feature to PXE less initialization, same as we have in PXE initialization.
Lukas, thoughts on this one?
Roman I can indeed implement the fix but I cannot reproduce, can you send me full logs from the discovered node when you encounter this kind of error? The script is called discovery-debug, send me output of this to the BZ.
I am asking because I don't understand, in PXE-less mode there is plenty of time if you go screen by screen. Do you mean semi-automatic or full-automatic (unattended) mode when you enter all the details on the kernel command line? What was your setting then? I need to know before I can put the delay.
(In reply to Lukas Zapletal from comment #3) > I am asking because I don't understand, in PXE-less mode there is plenty of > time if you go screen by screen. Do you mean semi-automatic or > full-automatic (unattended) mode when you enter all the details on the > kernel command line? What was your setting then? I need to know before I can > put the delay. I have asked for the logs, but the customer unfortunately does not have the HW to reproduce this issue. Have asked him for more details about booting mode and settings. Waiting for the reply.
The error message from the attached image is triggered by this statement: nmcli connection down primary this happens when the primary connection is brought down and it was not yet brought up. So the error is expected in this case. It's just a cosmetic issue and it has no effect. I filed this upstream: http://projects.theforeman.org/issues/19950 We will eventually fix this, no backport or QA needed for this one, closing.
Sorry, this was a different issue. Re-opening: http://projects.theforeman.org/issues/19984
Moving this bug to POST for triage into Satellite 6 since the upstream issue http://projects.theforeman.org/issues/19984 has been resolved.
Verified with foreman-discovery-image-3.4.1-3.iso. There is no clear reproducer for the issue so I did the following: 1) Used FDI to discover a host which worked correctly 2) Verified the change was actually present on the ISO in /usr/lib64/ruby/vendor_ruby/discovery/menu.rb 3) The error "nmcli connection down" is not present in journalctl Note: waiting for a magic constant "10 seconds" is fishy, it would be better to wait in a loop
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2018:0336