tests: Prefer permanent MAC for finding link info, fallback to current MAC #751

liangwen12year · 2025-01-18T21:36:50Z

Add the integration test tests_mac_address_match.yml to verify that the commit "Prioritize find link info by permanent MAC address, with fallback to current address" (c341683) can properly resolve the scenarios where multiple network interfaces share the same current MAC address, leading to potential ambiguity in link matching (for example, Virtual interfaces or VLANs, which often lack a valid perm-address and rely on the parent interface's address).

Notice that the test tests_mac_address_match.yml will be skipped in upstream testing and it will only be manually tested downstream since it requires to know the mac address to match when configuring vlan's parent.

Enhancement:

Reason:

Result:

Issue Tracker Tickets (Jira or BZ if any):

Resolves: https://issues.redhat.com/browse/RHEL-74211

codecov · 2025-01-18T21:37:57Z

Codecov Report

Attention: Patch coverage is 12.90323% with 27 lines in your changes missing coverage. Please review.

Project coverage is 43.03%. Comparing base (1b57520) to head (82ac075).
Report is 2 commits behind head on main.

Files with missing lines	Patch %	Lines
library/network_connections.py	12.90%	27 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #751      +/-   ##
==========================================
- Coverage   43.11%   43.03%   -0.09%     
==========================================
  Files          12       12              
  Lines        3124     3137      +13     
==========================================
+ Hits         1347     1350       +3     
- Misses       1777     1787      +10

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

tests/playbooks/tests_mac_address_match.yml

richm · 2025-01-18T22:46:26Z

tests/playbooks/tests_mac_address_match.yml

+    lsr_fail_debug:
+      - __network_connections_result
+  tags:
+    - "tests::mac"


@spetrosi I think we'll need to exclude this test from automated testing - we'll need to add tests::mac to the --skip-tags list @liangwen12year I think this test is meant to be run interactively

Created a PR for that linux-system-roles/tft-tests#65

I don't follow why the test cannot determine the MAC address itself instead of asking for it in a prompt.

The test is supposed to operate on an interface which has the permanent address, I do not want to make the test accidentally configured on the interface that is used to bridge the ssh connection between the control node and managed host. Possibly, I can search through all the links in /sys/class/net/ and use the first unconfigured ethernet interface. But such an ethernet interface (unconfigured and also have a permanent mac address) may not exist in the system environment of upstream testing.

You are right. We can use ethtool -P $iface to determine the mac address of the interface.

We controll the upstream test envirionment, so it should be clear whether or not such an interface exists and if it does not, there might be an option to create such an interface.

Nevertheless, the prompt also does not specify the requirements for the interface to ensure that it is eligible/the test does not ensure that it actually tests what it is supposed to test.

Honestly, it turns out that the upstream testing environment does not have the interface with the permanent mac address. I also tried iproute2 , systemd-networkd, and NetworkManager, but I can not create such an interface with the permanent mac address (the permanent mac address can be queried by calling ethtool -P $iface ), I can only create an interface with the current mac address (the current mac address can be queried by calling cat /sys/class/net/$iface/address ).

I specified the requirements for the interface to ensure that it is eligible.

How does the upstream environment work?

Summarized from the slack discussion, the upstream testing uses Testing Farm, which uses AWS machines, we do not run kvm/qemu virtual machines on the AWS machines. According to the doc https://tmt.readthedocs.io/en/stable/spec/hardware.html#network, requesting a machine with extra NICs should be already supported via network.type, we should be able to provision a machine with 2 NICs by specifying the following in the network system role test configuration.

hardware: network: - type: eth - type: eth

linux-system-roles/network#751 adds a test that must be skipped upstream and will only be tested downstream

tests/playbooks/tests_mac_address_match.yml

tyll · 2025-01-28T12:39:46Z

tests/playbooks/tests_mac_address_match.yml

+  hosts: all
+  vars_prompt:
+    - name: "interface"
+      prompt: "The test requires matching an Ethernet interface to its


From what I understand, the current MAC address of this interface needs to be different than the permanent MAC address for the new code to actually be tested. And after further discussion, there should probably be a second interface (maybe a virtual one) that has the same MAC address as the permanent MAC address as this was causing an error condition. @liangwen12year told me about this in a meeting, AFAIU, this is not mentioned in the original PR.

https://issues.redhat.com/browse/RHEL-73442, the issue occurs when both interface name and MAC address are specified in network connections to configure a parent and VLAN connection, and the configuration is applied multiple times. The error is raised because SysUtil._link_infos_fetch() creates a dictionary of link info from /sys/class/net/, comparing the user-specified MAC address against either the "perm-address" or "address." If the user-specified MAC matches only the "address" (current MAC) but not the "perm-address" (permanent MAC), the link info returned may have a different interface name than specified, leading to the error. If a physical interface has permanent MAC A and current MAC B, specifying MAC A in the network connection avoids errors because the VLAN's current MAC matches the parent's, but the VLAN’s perm-address is unset, so its link info isn’t used for comparing the interface name against the user-specified one. However, specifying MAC B causes NetworkManager to fail, as it expects the interface name to match the permanent MAC address, not the current one.

From my understanding, the current mac address of this interface should be the same as the permanent mac, then the created vlan interface will have the same current mac address as the permanent mac address of its parent (a virtual interface like a VLAN typically inherits the current MAC address of its parent interface instead of the permanent MAC.), then without the prior fix (c341683) on the issue, I can reproduce the problem by applying the same network connections multiple times for configuring the parent and the vlan connections.

library/network_connections.py

liangwen12year · 2025-02-12T20:23:57Z

[citest]

liangwen12year · 2025-02-17T15:09:53Z

[citest]

liangwen12year · 2025-03-10T22:02:29Z

[citest]

liangwen12year · 2025-03-11T00:06:41Z

[citest]

liangwen12year · 2025-03-11T17:12:48Z

[citest]

liangwen12year · 2025-03-12T02:03:37Z

[citest]

The existing implementation of `link_info_find()` had several issues that made it error-prone and harder to maintain: - The refresh parameter was unused, adding unnecessary complexity. - MAC address matching logic was convoluted, with redundant variables and conditions. - The use of locals() for fallback results was unclear and could lead to subtle bugs. - The order of checks (MAC vs ifname) didn't prioritize the most reliable matching criteria. To address these issues, the following changes were made: - Removed the unused `refresh` argument to reduce complexity. - Used `None` checks more idiomatically with `if mac` instead of `is not None`. - Eliminated redundant variables and conditions to improve readability. - Avoided using `locals()` by explicitly storing fallback results. - Made `ifname` matching take priority before checking MAC addresses. - Ensured that the function returns early when a definitive match is found. Signed-off-by: Wen Liang <[email protected]>

When a user provides both an interface name and a MAC address, the current validation process retrieves sysfs link info separately using the interface name and the MAC address, then compares the results. If the information doesn't match, an error is raised. However, this approach may trigger false alarms because retrieving the link info by MAC might return data that only matches the current MAC instead of the permanent MAC. Since the interface name is unique within the kernel, a more robust validation method is to fetch the MAC address using the interface name and then compare it directly with the user-provided MAC address. Signed-off-by: Wen Liang <[email protected]>

…t MAC When a network connection specifies both an interface name and MAC address, and the physical interface has identical permanent and current MACs, applying the configuration multiple times can cause an error: "no such interface exists." The issue occurs because `SysUtil.link_info_find()` may return a link info entry where the user-specified MAC matches only the current ("address") but not the permanent MAC ("perm-address"). In such cases, the returned link info may have a different interface name than the one specified in the network connection, leading to the error. We already implemented a fix (c341683) that ensures link lookup prioritizes the permanent MAC, only falling back to the current MAC when necessary. The integration test `tests_mac_address_match.yml` verifies this fix by requiring an Ethernet interface where the permanent and current MACs match. Resolves: https://issues.redhat.com/browse/RHEL-74211 Signed-off-by: Wen Liang <[email protected]>

liangwen12year · 2025-03-12T18:05:27Z

[citest]

liangwen12year · 2025-03-13T23:00:21Z

Close this PR as the patches in this PR have been split into multiple separate PRs.

liangwen12year force-pushed the link_info_find_fix branch 2 times, most recently from 75a9f57 to 855766f Compare January 18, 2025 21:39

richm reviewed Jan 18, 2025

View reviewed changes

tests/playbooks/tests_mac_address_match.yml Outdated Show resolved Hide resolved

richm reviewed Jan 18, 2025

View reviewed changes

tests/playbooks/tests_mac_address_match.yml Outdated Show resolved Hide resolved

richm reviewed Jan 18, 2025

View reviewed changes

liangwen12year force-pushed the link_info_find_fix branch from 855766f to 4538327 Compare January 19, 2025 00:32

spetrosi added a commit to spetrosi/tft-tests that referenced this pull request Jan 20, 2025

Add tests::mac to the list of skipped tags

6ab4ed7

linux-system-roles/network#751 adds a test that must be skipped upstream and will only be tested downstream

spetrosi mentioned this pull request Jan 20, 2025

Add tests::mac to the list of skipped tags linux-system-roles/tft-tests#65

Merged

liangwen12year requested review from cathay4t and ffmancera January 21, 2025 16:09

tyll reviewed Jan 21, 2025

View reviewed changes

tests/playbooks/tests_mac_address_match.yml Show resolved Hide resolved

liangwen12year force-pushed the link_info_find_fix branch 9 times, most recently from a1bac2e to 9cd64bb Compare January 28, 2025 04:08

tyll reviewed Jan 28, 2025

View reviewed changes

liangwen12year force-pushed the link_info_find_fix branch 3 times, most recently from 50594c8 to c4ac501 Compare February 4, 2025 14:15

richm reviewed Feb 4, 2025

View reviewed changes

library/network_connections.py Outdated Show resolved Hide resolved

richm reviewed Feb 4, 2025

View reviewed changes

library/network_connections.py Show resolved Hide resolved

liangwen12year force-pushed the link_info_find_fix branch from c4ac501 to 1871ed1 Compare February 4, 2025 21:07

tyll reviewed Feb 5, 2025

View reviewed changes

library/network_connections.py Outdated Show resolved Hide resolved

tyll reviewed Feb 5, 2025

View reviewed changes

library/network_connections.py Show resolved Hide resolved

liangwen12year force-pushed the link_info_find_fix branch from f63c768 to a380ac4 Compare February 12, 2025 20:05

liangwen12year force-pushed the link_info_find_fix branch from a380ac4 to 052fc9b Compare February 17, 2025 15:09

liangwen12year changed the title ~~tests: Prioritize find link info by permanent MAC address, with fallback to current address~~ fix: Ensure deterministic sysfs link info fetching, refactor link_info_find, and add MAC address match integration tests Feb 18, 2025

liangwen12year changed the title ~~fix: Ensure deterministic sysfs link info fetching, refactor link_info_find, and add MAC address match integration tests~~ fix: Ensure deterministic sysfs link info fetching, refactor link_info_find Feb 18, 2025

liangwen12year changed the title ~~fix: Ensure deterministic sysfs link info fetching, refactor link_info_find~~ fix: Ensure deterministic sysfs link info fetching abd refactor link_info_find Feb 18, 2025

liangwen12year changed the title ~~fix: Ensure deterministic sysfs link info fetching abd refactor link_info_find~~ fix: Ensure deterministic sysfs link info fetching and refactor link_info_find Feb 18, 2025

liangwen12year force-pushed the link_info_find_fix branch 2 times, most recently from ce44faf to 57b4c00 Compare February 25, 2025 07:15

liangwen12year force-pushed the link_info_find_fix branch 2 times, most recently from a136ccd to 869e15e Compare March 10, 2025 22:00

liangwen12year force-pushed the link_info_find_fix branch from 869e15e to b1f1407 Compare March 11, 2025 00:06

liangwen12year force-pushed the link_info_find_fix branch from b1f1407 to 6325901 Compare March 11, 2025 17:08

liangwen12year force-pushed the link_info_find_fix branch from 6325901 to 7d316a7 Compare March 12, 2025 02:00

liangwen12year force-pushed the link_info_find_fix branch 2 times, most recently from 1c3b84c to 64a5f07 Compare March 12, 2025 14:24

liangwen12year added the Wait_Other_PR label Mar 12, 2025

liangwen12year changed the title ~~fix: Ensure deterministic sysfs link info fetching and refactor link_info_find~~ tests: Prefer permanent MAC for finding link info, fallback to current MAC Mar 12, 2025

liangwen12year added 3 commits March 12, 2025 13:57

liangwen12year force-pushed the link_info_find_fix branch from 64a5f07 to 82ac075 Compare March 12, 2025 18:02

liangwen12year closed this Mar 13, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

tests: Prefer permanent MAC for finding link info, fallback to current MAC #751

tests: Prefer permanent MAC for finding link info, fallback to current MAC #751

liangwen12year commented Jan 18, 2025 •

edited

Loading

codecov bot commented Jan 18, 2025 •

edited

Loading

richm Jan 18, 2025

spetrosi Jan 20, 2025

tyll Jan 21, 2025

liangwen12year Jan 21, 2025 •

edited

Loading

liangwen12year Jan 22, 2025

tyll Jan 22, 2025

liangwen12year Jan 25, 2025

liangwen12year Jan 27, 2025

tyll Jan 27, 2025

liangwen12year Feb 3, 2025

tyll Jan 28, 2025

liangwen12year Jan 28, 2025 •

edited

Loading

liangwen12year commented Feb 12, 2025

liangwen12year commented Feb 17, 2025

liangwen12year commented Mar 10, 2025

liangwen12year commented Mar 11, 2025

liangwen12year commented Mar 11, 2025

liangwen12year commented Mar 12, 2025

liangwen12year commented Mar 12, 2025

liangwen12year commented Mar 13, 2025 •

edited

Loading

tests: Prefer permanent MAC for finding link info, fallback to current MAC #751

tests: Prefer permanent MAC for finding link info, fallback to current MAC #751

Conversation

liangwen12year commented Jan 18, 2025 • edited Loading

codecov bot commented Jan 18, 2025 • edited Loading

Codecov Report

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

liangwen12year Jan 21, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

liangwen12year Jan 28, 2025 • edited Loading

Choose a reason for hiding this comment

liangwen12year commented Feb 12, 2025

liangwen12year commented Feb 17, 2025

liangwen12year commented Mar 10, 2025

liangwen12year commented Mar 11, 2025

liangwen12year commented Mar 11, 2025

liangwen12year commented Mar 12, 2025

liangwen12year commented Mar 12, 2025

liangwen12year commented Mar 13, 2025 • edited Loading

liangwen12year commented Jan 18, 2025 •

edited

Loading

codecov bot commented Jan 18, 2025 •

edited

Loading

liangwen12year Jan 21, 2025 •

edited

Loading

liangwen12year Jan 28, 2025 •

edited

Loading

liangwen12year commented Mar 13, 2025 •

edited

Loading