Summary: | sys-apps/openrc-0.9.8.1 breaks network bonding | ||
---|---|---|---|
Product: | Gentoo Hosted Projects | Reporter: | Michael Reithinger <mreithinger> |
Component: | OpenRC | Assignee: | OpenRC Team <openrc> |
Status: | RESOLVED FIXED | ||
Severity: | major | CC: | atoth, bircoph |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
Whiteboard: | openrc:oldnet | ||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 399185, 400787 | ||
Attachments: | And here is the fix |
Description
Michael Reithinger
2012-01-24 17:25:21 UTC
The same problem here with openrc-0.9.8.1: in raises bond-dependent vlans not on bond0, but on the last bond's slave (eth1): vlan23@eth1, vlan25@eth1, etc instead of vlan23@bond0, vlan25@bond0, etc... when vlan_bond0="23 25" slaves_bond0="eth0 eth1" Sorry, can't post more ATM, because I lost networks access to my system due to this nice, yummy, tasty, wonderful, ingenious update and I'm working with it using ilo. The same shit with 0.9.8, version 0.9.7 worked well before update. But now, so thanks to someone hastily removed 0.9.7 from the tree, I have to revert to 0.9.4 8-( of course, it also works. Some problem here: XXX ~ # cat /etc/conf.d/net config_eth0="null" config_eth1="null" slaves_bond0="eth0 eth1" dns_servers_bond0="172.17.1.1" config_bond0="172.17.1.130/24" routes_bond0="default gw 172.17.1.20" After downgrade do 0.9.4 all working good. Created attachment 299817 [details, diff]
And here is the fix
The bonding script does a lot of funny stuff, including setting the slaves as IFACE one by one and running commands on them preparing them for the bond.
However currently OpenRC does not unset IFACE (and thus resetting it to the name of the bond) before it returns, making the last slave it poke at the current IFACE, resulting in the rest of the network setup being ran against that slave instead of the bond.
One line fix provided.
Also: tested on a stable system. Has run it on a ~arch system before. Got bitten by this in a former upgrade, and now again this night after I upgraded to OpenRC-0.9.8.1, so it annoyed me enough to post the patch. Please fix soon. Fixed in 0.9.8.2, will be rapid-stable *** Bug 400847 has been marked as a duplicate of this bug. *** (In reply to comment #5) > Also: tested on a stable system. Has run it on a ~arch system before. Got > bitten by this in a former upgrade, and now again this night after I upgraded > to OpenRC-0.9.8.1, so it annoyed me enough to post the patch. > > Please fix soon. Hi Xake! Thanks for the fix! Dw. |