Bug 1908462
| Summary: | RHCOS 4.6 ostree removed dhclient | |||
|---|---|---|---|---|
| Product: | OpenShift Container Platform | Reporter: | Jonathan Edwards <joedward> | |
| Component: | RHCOS | Assignee: | Christian Glombek <cglombek> | |
| Status: | CLOSED ERRATA | QA Contact: | Michael Nguyen <mnguyen> | |
| Severity: | urgent | Docs Contact: | ||
| Priority: | medium | |||
| Version: | 4.6.z | CC: | antgarci, apaladug, bbreard, bgalvani, dmoessne, dornelas, dustymabe, imcleod, jligon, miabbott, nstielau, smilner, walters, wzheng | |
| Target Milestone: | --- | Keywords: | UpcomingSprint | |
| Target Release: | 4.7.0 | |||
| Hardware: | Unspecified | |||
| OS: | Unspecified | |||
| Whiteboard: | ||||
| Fixed In Version: | Doc Type: | Bug Fix | ||
| Doc Text: |
Cause: The dhclient binary was removed from RHCOS when the switch to using NetworkManager in the initramfs was made.
Consequence: Users that had configured RHCOS to use dhclient for DHCP operations were left with systems that were unable to properly acquire a DHCP address.
Fix: Include the dhclient binary in RHCOS.
Result: RHCOS systems are able to perform DHCP operations using dhclient successfully.
|
Story Points: | --- | |
| Clone Of: | ||||
| : | 1915997 (view as bug list) | Environment: | ||
| Last Closed: | 2021-02-24 15:45:31 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: | ||||
| Bug Depends On: | ||||
| Bug Blocks: | 1915997, 1922292 | |||
|
Description
Jonathan Edwards
2020-12-16 18:53:13 UTC
Hey Jonathan, Any chance we could gain some insight into what use cases are missing that make NetworkManager's internal DHCP agent insufficient? We'd like to follow up on these with the NetworkManager team. There's a few corner cases where we're still finding dhclient succeeds where the internal NetworkManager dhcp client fails. Given the pervasiveness of dhclient over the years, it's desirable to have it as a fallback in case anything goes awry in a customer environment where dhclient works and the internal dhcp fails. > What do you suggest as the next steps to investigate the issue? It seems to me this is the same issue as: https://bugzilla.redhat.com/show_bug.cgi?id=1883521#c52 https://github.com/openshift/openshift-docs/pull/26455 What is the content of /etc/dhcp/dhclient.conf on 4.5? This issue has been discussed with PM, CEE, and Engineering and we've concluded that the best path forward is to provide the `dhclient` binary in RHCOS for 4.6 and 4.7. We will include a deprecation notice as part of the OCP 4.7 release notes, as we would like to remove the binary from RHCOS as part of OCP 4.8. Note: this plan does not cover including the `dhclient` binary in the RHCOS initrd for 4.6 or 4.7. However, we will continue to default to using the internal NetworkManager DHCP client in 4.6, 4.7 and future releases. This means that customers that wish to continue to use `dhclient` for their DHCP needs in 4.6 and 4.7, will need to provide a config file in `/etc/NetworkManager/conf.d` that instructs NetworkManager to use `dhclient`, *before* upgrading to 4.6. For example: ``` $ cat /etc/NetworkManager/conf.d/dhcp-client.conf [main] dhcp=dhclient ``` This can be done with a properly formatted MachineConfig. We've speculated that an incorrectly configured `dhclient.conf` file could explain why DHCP requests in RHCOS 4.5 were using something other than the MAC address as the DHCP client ID. When the upgrade to 4.6 happened, the default behavior of the internal NetworkManager DHCP client uses the MAC address and therefore a different IP was assigned. However, without the entirety of any `dhclient.conf` on the nodes and/or config files under `/etc/NetworkManager` it is purely speculation. As part of fixing and verifying this BZ, we would need a proper reproducer that we can execute to prove that the addition of the `dhclient` binary will solve the issue during the upgrade from RHCOS 4.5 to RHCOS 4.6 (with `dhclient` binary). @Jonathan is this something that you could help us with? Lastly, this BZ should be used to track the inclusion of the `dhclient` binary in RHCOS 4.7; I will clone this BZ to track the inclusion of the `dhclient` binary in RHCOS 4.6.z Higher priority work has prevented from this issue being worked on; adding UpcomingSprint keyword Checked 47.83.202101301239-0 which is the latest referenced ART build. [core@cosa-devsh ~]$ ls /usr/sbin/dhclient /usr/sbin/dhclient-script ls: cannot access '/usr/sbin/dhclient': No such file or directory ls: cannot access '/usr/sbin/dhclient-script': No such file or directory [core@cosa-devsh ~]$ However, checking the latest prow passing image in CI (47.83.202102031754-0) I do see the change: [core@cosa-devsh ~]$ ls /usr/sbin/dhclient /usr/sbin/dhclient-script /usr/sbin/dhclient /usr/sbin/dhclient-script [core@cosa-devsh ~]$ Marking as verified. 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 (Moderate: OpenShift Container Platform 4.7.0 security, 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/RHSA-2020:5633 |