Added pciid match support#375
Conversation
Signed-off-by: Fatih USTA <fatihusta86@gmail.com>
Signed-off-by: Fatih USTA <fatihusta86@gmail.com>
Signed-off-by: Fatih USTA <fatihusta86@gmail.com>
slyon
added
the
schema change
|
Sorry, this is taking much longer than anticipated. I like the idea of having |
slyon
added
the
community
slyon
left a comment
slyon
left a comment
There was a problem hiding this comment.
My apologies, this PR fell through the cracks. I've now done an initial high-level round of review, as this is touching quite some core parts of Netplan and the original issue seems to be fixed as of recently (#503).
Please let me know if you're still interested in driving this forward, so I can do a more detailed review and talk to our architect about the new YAML schema.
|
|
||
| NETPLAN_PUBLIC gboolean | ||
| netplan_netdef_match_interface(const NetplanNetDefinition* netdef, const char* name, const char* mac, const char* driver_name); | ||
| netplan_netdef_match_interface(const NetplanNetDefinition* netdef, const char* name, const char* mac, const char* driver_name, const char* pciid); |
There was a problem hiding this comment.
todo: We cannot just change public API like this, but would rather need to introduce a new netplan_netdef_match_interface2() function. Keeping the original function backwards compatible.
| char* driver; | ||
| char* mac; | ||
| char* original_name; | ||
| char* pciid; |
There was a problem hiding this comment.
todo: This has the potential of breaking ABI and we'd need to move the new field to the very end of the netplan_net_definition struct, e.g. called match_pciid.
| - **pciid** (scalar) – since **0.107** | ||
|
|
||
| > The PCI ID, corresponding to the `ID_PATH` udev properity. (`0000:98:00.1`) | ||
| > Matching on pci id is *only* supported with networkd. | ||
| > `udevadm info /sys/class/net/DEVICE_NAME` |
There was a problem hiding this comment.
nitpick (non-blocking): Version needs to be bumped.
question: Also, is there any possibility of doing the same for the NetworkManager backend? The match stanza is a very central piece of Netplan's configuration and we should try not to diverge between the backends here.
| {"driver", YAML_NO_NODE, {.variable=handle_match_driver}}, | ||
| {"macaddress", YAML_SCALAR_NODE, {.generic=handle_netdef_mac}, netdef_offset(match.mac)}, | ||
| {"name", YAML_SCALAR_NODE, {.generic=handle_netdef_id}, netdef_offset(match.original_name)}, | ||
| {"pciid", YAML_SCALAR_NODE, {.generic=handle_netdef_str}, netdef_offset(match.pciid)}, |
There was a problem hiding this comment.
todo: We'd probably want to have a more specific handler here, that validates the input string is an actual PCI ID.
| self.assert_nm(None) | ||
| self.assert_nm_udev(NM_UNMANAGED % 'lom1' + NM_UNMANAGED_MAC % '11:22:33:44:55:66') | ||
|
|
||
| def test_eth_match_by_pciid_rename(self): |
There was a problem hiding this comment.
nitpick: Thanks for providing some test cases! I'd love to see an integration test along the lines of tests/integration/ethernets.py:test_rename_interfaces(), too (if possible).
| return link.get('addr', '') | ||
|
|
||
|
|
||
| def get_interface_pciid(interface): # pragma: nocover (covered in autopkgtest) |
There was a problem hiding this comment.
question: Can we apply some test-mocking to get this net method covered by unit tests?
2 participants
Description
Netplan doesn't match mac address of ethernet when using bond. So member ethernet doesn't up.
This PR fixes ethernet up problem, but still prints warning message(below).
I try to add PCI ID match support for better renaming to the ethernet.
Checklist
make checksuccessfully.make check-coverage).Also may fixes this issue: https://bugs.launchpad.net/netplan/+bug/1608447