Bug 1853277
| Summary: | anaconda sets ONBOOT=yes for NICs not used for installation | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Red Hat Enterprise Linux 8 | Reporter: | Jan Tluka <jtluka> | ||||
| Component: | NetworkManager | Assignee: | Beniamino Galvani <bgalvani> | ||||
| Status: | CLOSED ERRATA | QA Contact: | Desktop QE <desktop-qa-list> | ||||
| Severity: | unspecified | Docs Contact: | |||||
| Priority: | unspecified | ||||||
| Version: | 8.3 | CC: | acardace, atragler, bgalvani, lnst-team, lnykryn, lrintel, rkhan, rvykydal, sukulkar, thaller, till, vbenes | ||||
| Target Milestone: | rc | Keywords: | Regression, TestBlocker | ||||
| Target Release: | 8.3 | Flags: | pm-rhel:
mirror+
|
||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Whiteboard: | |||||||
| Fixed In Version: | NetworkManager-1.26.0-0.2.1.el8 | Doc Type: | If docs needed, set a value | ||||
| Doc Text: | Story Points: | --- | |||||
| Clone Of: | Environment: | ||||||
| Last Closed: | 2020-11-04 01:50:15 UTC | Type: | Bug | ||||
| Regression: | --- | Mount Type: | --- | ||||
| Documentation: | --- | CRM: | |||||
| Verified Versions: | Category: | --- | |||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||
| Embargoed: | |||||||
| Attachments: |
|
||||||
|
Description
Jan Tluka
2020-07-02 10:43:47 UTC
Jan, could you please attach logs from the installation? Either /tmp/syslog from installed environment or /var/log/anaconda/journal.log from installed system. (In reply to Radek Vykydal from comment #1) > Jan, could you please attach logs from the installation? > Either /tmp/syslog from installed environment or I mean from *installer* environment I am not able to reproduce the issue in common use cases, also our kickstart tests don't indicate any regression. Created attachment 1699632 [details]
anaconda logs RHEL-8.3.0-20200616.0
Logs from the installer, compose RHEL-8.3.0-20200616.0.
(In reply to Jan Tluka from comment #4) > Created attachment 1699632 [details] > anaconda logs RHEL-8.3.0-20200616.0 > > Logs from the installer, compose RHEL-8.3.0-20200616.0. Relevant devices are em1 and em2. For the reference, here's a beaker job link. https://beaker.engineering.redhat.com/jobs/4396507 (In reply to Jan Tluka from comment #5) > (In reply to Jan Tluka from comment #4) > > Created attachment 1699632 [details] > > anaconda logs RHEL-8.3.0-20200616.0 As per IRC communication with Jan the issue started to appear at this compose, which could point to a new NetworkManager version (1:1.25.2-1 -> 1:1.26.0-0.1). There are no changes in Anaconda that could cause a regression of this kind (anconda 33.16.3.5-1 -> 33.16.3.6-1). I think a change in NM could maybe cause the regression by some race condition starting to appear, I am not not sure which rhel 8.3 compose (NM version) was actually working for the setup. Nevertheless, the root cause - and regression / change of behavior compared to RHEL 8.2 seems to be handling of BOOTIF installer boot option (supplied by PXE) by NM dracut module. > > Relevant devices are em1 and em2. Unlike in RHEL 8.2 (dracut network module) the em1 and em2 devices are unsuccessfully tried to be activated in initramfs (using dhcp) and also after switchroot, which is making the ONBOOT value to be set to yes for installed system (assuming there was an intention to have them activated/configured in installer and therefore in target system). In RHEL 8.2 only the device specified by BOOTIF option (em3) would be activated in intramfs. Thomas, any ideas about NM dracut module honouring BOOTIF option (as the legacy dracut network module used to I believe) ? I'll look at it as well. (In reply to Radek Vykydal from comment #7) > Thomas, any ideas about NM dracut module honouring BOOTIF option (as the > legacy dracut network module used to I believe) ? Asking also Lukas from dracut. > Thomas, any ideas about NM dracut module honouring BOOTIF option (as the > legacy dracut network module used to I believe) ? > I'll look at it as well. Connections generated in initramfs: # cat /run/NetworkManager/system-connections/default_connection.nmconn [connection] id=Wired Connection uuid=58e90d13-6186-470e-b8d3-480899ab1489 type=ethernet multi-connect=3 permissions= [ethernet] mac-address-blacklist= [ipv4] dns-search= method=auto [ipv6] addr-gen-mode=eui64 dns-search= method=auto [proxy] pre-pivot:/# cat /proc/cmdline initrd=test/rv/scripted/r83-latest/initrd.img repo=http://download.eng.brq.redhat.com/rhel-8/nightly/RHEL-8/latest-RHEL-8.3.0/compose/BaseOS/x86_64/os/ inst.addrepo=APPSTREAM,http://download.eng.brq.redhat.com/rhel-8/nightly/RHEL-8/latest-RHEL-8.3/compose/AppStream/x86_64/os/ updates=http://10.43.136.2/ks/rv/updates.bootvirtual.img ksdevice=52:54:00:9f:21:28 ks=http://10.43.136.2/ks/rv/ks.onboot83.cfg rd.break=pre-pivot rd.bootif=1 console=ttyS0 BOOT_IMAGE=test/rv/scripted/r83-latest/vmlinuz BOOTIF=01-52-54-00-9f-21-28 pre-pivot:/# /usr/libexec/nm-initrd-generator BOOTIF=01-52-54-00-9f-21-28 -s (nm-initrd-generator:1546): libnm-CRITICAL **: 11:20:48.010: ((libnm-core/nm-setting-wired.c:205)): assertion '<dropped>' failed (nm-initrd-generator:1546): GLib-GObject-CRITICAL **: 11:20:48.010: g_object_set: assertion 'G_IS_OBJECT (object)' failed *** Connection 'default_connection' *** [connection] id=Wired Connection uuid=b693a970-af74-47e3-ad69-eaab89fa7678 type=ethernet multi-connect=3 permissions= [ethernet] mac-address-blacklist= [ipv4] dns-search= method=auto [ipv6] addr-gen-mode=eui64 dns-search= method=auto [proxy] Interestingly, if I add ip=dhcp to the boot options the generated connection seems to be as some we would expect (bound to MAC): pre-pivot:/# /usr/libexec/nm-initrd-generator -s -- BOOTIF=01-52-54-00-9f-21-28 ip=dhcp *** Connection 'default_connection' *** [connection] id=Wired Connection uuid=0cb1ef31-a1c9-4541-9a5f-c5d87c4ac8be type=ethernet multi-connect=3 permissions= [ethernet] mac-address=52:54:99:9F:21:28 mac-address-blacklist= [ipv4] dns-search= may-fail=false method=auto [ipv6] addr-gen-mode=eui64 dns-search= method=auto [proxy] Hi, this is caused by a bug in the NM initrd generator, which creates a default connection without a mac-address property. This upstream merge request fixes it: https://gitlab.freedesktop.org/NetworkManager/NetworkManager/-/merge_requests/562 I think this bz can be reassigned to NM. Adding TestBlocker. There's no workaround other than to use RHEL-8.2.0 for our test runs because of this bug. For the reference, PR for Anaconda kickstart test for the issue: https://github.com/rhinstaller/kickstart-tests/pull/348 Covered in anaconda test suite. No auto test case added. Just default device now has ONBOOT set. 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 (NetworkManager bug fix and enhancement update), 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/RHBA-2020:4499 |