dracut, udev and complicated netboot cases
cover dracut, udev and complicated netboot cases
Hello all
The following two patches add udev driven network configuration support and a basic nfsroot handler.
As describe earlier, there can be many weird and complicated netboot cases. The requirement that it should be udev driven doesn't make it any easier. At least at first glance.
The current solution actually makes use of udev for speed up. The idea is based on Victor Lowther's first implementation which forked off dhclients after the udev event fired. By modifying ifup and the dhclient-script we can synchronize back by using a simple (blocking) FIFO. Interfaces which are ready are upped and tested FIFO-style by reading one out of the fifo, and handing over the the actual mount-script. If it succeeds, all is well. If it fails or times out, we try (or wait for) the next interface.
The whole interface configuration part tries to behave as much like the current implementation inside the kernel as possible, with a few extensions:
- I'm assuming that if there's a network driver inside the initrd the user wants something. So no ip= option is treated like ip=dhcp
- Instead of only relying on the first interface that pops up or the one that first receives a dhcp reply we try all interfaces that have a useable configuration.
Patches apply on top of Harald Hoyers merge tree
(git://git.surfsite.org/pub/git/dracut.git)
Regards,
Philippe
--
To unsubscribe from this list: send the line "unsubscribe initramfs" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
Quoted 18 lines · Seewer Philippe
Hello all
The following two patches add udev driven network configuration support and a basic nfsroot handler.
As describe earlier, there can be many weird and complicated netboot cases. The requirement that it should be udev driven doesn't make it any easier. At least at first glance.
The current solution actually makes use of udev for speed up. The idea is based on Victor Lowther's first implementation which forked off dhclients after the udev event fired. By modifying ifup and the dhclient-script we can synchronize back by using a simple (blocking) FIFO. Interfaces which are ready are upped and tested FIFO-style by reading one out of the fifo, and handing over the the actual mount-script. If it succeeds, all is well. If it fails or times out, we try (or wait for) the next interface.
Quoted 8 lines
The whole interface configuration part tries to behave as much like the current implementation inside the kernel as possible, with a few extensions:
- I'm assuming that if there's a network driver inside the initrd the user wants something. So no ip= option is treated like ip=dhcp
- Instead of only relying on the first interface that pops up or the one that first receives a dhcp reply we try all interfaces that have a useable configuration.
You shoudl save the additional dhcp options, so that people who want to do their fs mounting based on parameters provided by dhcp can do that.
Quoted 9 lines
Patches apply on top of Harald Hoyers merge tree
(git://git.surfsite.org/pub/git/dracut.git)Regards,
Philippe
--
To unsubscribe from this list: send the line "unsubscribe initramfs" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
Signature · 8 lines
Victor Lowther RHCE# 805008539634727 LPIC-2# LPI000140019 -- To unsubscribe from this list: send the line "unsubscribe initramfs" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
Quoted 2 lines · Victor Lowther
You shoudl save the additional dhcp options, so that people who want to do their fs mounting based on parameters provided by dhcp can do that.
Good point. Save it in a sourceable scriptfile or just let dhclient create a leasefile?
--
To unsubscribe from this list: send the line "unsubscribe initramfs" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html