I'm not sure if this is the right place to file it, but I've encountered quite a conundrum with the new EC2 CentOS CI machines.
After a week of going back and forth between the EC2 VMs and metal machines, I finally found out a suitable middle-ground for our systemd tests by using metal machines, but skipping the costly reboot (which currently takes 10 - 20 minutes).
I've got almost everything working except one part - our systemd-networkd test suite needs extra kernel modules from the kernel-modules-extra package. However, the kernel which the metal machines currently run is quite old - so old, that it's not even on mirrors anymore, including the module packages:
kernel-modules-extra
# uname -r 4.18.0-305.3.1.el8.x86_64
Packages on one of the C8S mirrors:
[ ] kernel-4.18.0-305.10.2.el8_4.x86_64.rpm 2021-07-20 21:12 5.9M [ ] kernel-4.18.0-305.12.1.el8_4.x86_64.rpm 2021-08-11 05:46 5.9M [ ] kernel-4.18.0-305.17.1.el8_4.x86_64.rpm 2021-09-08 17:47 5.9M [ ] kernel-4.18.0-305.19.1.el8_4.x86_64.rpm 2021-09-15 20:50 5.9M [ ] kernel-4.18.0-305.25.1.el8_4.x86_64.rpm 2021-11-03 12:55 5.9M [ ] kernel-4.18.0-315.el8.x86_64.rpm 2021-06-28 22:45 6.4M [ ] kernel-4.18.0-326.el8.x86_64.rpm 2021-07-29 00:48 6.8M [ ] kernel-4.18.0-331.el8.x86_64.rpm 2021-08-19 20:23 6.9M [ ] kernel-4.18.0-338.el8.x86_64.rpm 2021-08-27 20:43 7.0M [ ] kernel-4.18.0-348.2.1.el8_5.x86_64.rpm 2021-11-17 17:09 7.0M [ ] kernel-4.18.0-348.7.1.el8_5.x86_64.rpm 2021-12-22 15:51 7.0M [ ] kernel-4.18.0-348.el8.x86_64.rpm 2021-10-20 18:52 7.0M [ ] kernel-4.18.0-358.el8.x86_64.rpm 2022-01-10 16:54 7.5M [ ] kernel-4.18.0-365.el8.x86_64.rpm 2022-02-10 18:54 7.7M [ ] kernel-4.18.0-373.el8.x86_64.rpm 2022-03-22 18:34 8.0M [ ] kernel-4.18.0-383.el8.x86_64.rpm 2022-04-20 22:10 8.1M [ ] kernel-4.18.0-394.el8.x86_64.rpm 2022-06-03 16:54 8.3M [ ] kernel-4.18.0-408.el8.x86_64.rpm 2022-07-19 16:50 8.5M
Link: https://mirror-prg.webglobe.com/centos/8-stream/BaseOS/x86_64/os/Packages/
Now, cloud-init does a kernel update when the machine is provisioned, but the machine is not rebooted afterwards, so the runtime kernel remains the same:
cloud-init
# journalctl --no-hostname -o short-monotonic -b _PID=2655 | grep kernel [ 85.847586] cloud-init[2655]: kernel x86_64 4.18.0-408.el8 baseos 8.5 M [ 85.847586] cloud-init[2655]: kernel-core x86_64 4.18.0-408.el8 baseos 40 M [ 85.847586] cloud-init[2655]: kernel-modules x86_64 4.18.0-408.el8 baseos 32 M [ 85.862512] cloud-init[2655]: kernel-tools x86_64 4.18.0-408.el8 baseos 8.8 M [ 85.862512] cloud-init[2655]: kernel-tools-libs x86_64 4.18.0-408.el8 baseos 8.5 M [ 86.625977] cloud-init[2655]: (5/266): kernel-4.18.0-408.el8.x86_64.rpm 25 MB/s | 8.5 MB 00:00 [ 87.568341] cloud-init[2655]: (7/266): kernel-core-4.18.0-408.el8.x86_64.rpm 33 MB/s | 40 MB 00:01 [ 88.204240] cloud-init[2655]: (10/266): kernel-modules-4.18.0-408.el8.x86_64. 18 MB/s | 32 MB 00:01 [ 95.083312] cloud-init[2655]: (118/266): kernel-tools-libs-4.18.0-408.el8.x86 32 MB/s | 8.5 MB 00:00 [ 95.236929] cloud-init[2655]: (119/266): kernel-tools-4.18.0-408.el8.x86_64.r 21 MB/s | 8.8 MB 00:00 ...
Aren't the CentOS AMIs supposed to be updated regularly, especially since C8S is now basically a rolling release?
This should be opened on Red Hat Bugzilla as CentOS Stream ticket.
Metadata Update from @zlopez: - Issue close_status updated to: Wrong tracker - Issue status updated to: Closed (was: Open)
FTR, reported in RHBZ as https://bugzilla.redhat.com/show_bug.cgi?id=2116386.
Log in to comment on this ticket.