Please revert commit "There is only one resolv.conf."
Having only one resolv.conf is ok for a booted client. That is the normal and the usual behaviour is that the first interface to be up and ready with data is the one to write the contents.
But inside dracut we really do not want that behaviour. First the primary interface (see bootdev cmdline argument) might not be the first interface to come up, second if we are in guess-the-interface mode, multiple interface might receive different data for resolv.conf and we have to use the correct stuff to mount.
So please revert that commit.
Thank you,
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 11 lines · Seewer Philippe
Having only one resolv.conf is ok for a booted client. That is the normal and the usual behaviour is that the first interface to be up and ready with data is the one to write the contents.
But inside dracut we really do not want that behaviour. First the primary interface (see bootdev cmdline argument) might not be the first interface to come up, second if we are in guess-the-interface mode, multiple interface might receive different data for resolv.conf and we have to use the correct stuff to mount.
So please revert that commit.
Do you agree the one copied ultimately should be the interface that successfully netbooted?
Warren
--
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 11 lines · Warren Togami
Quoted 10 lines · Seewer Philippe
Having only one resolv.conf is ok for a booted client. That is the normal and the usual behaviour is that the first interface to be up and ready with data is the one to write the contents.
But inside dracut we really do not want that behaviour. First the primary interface (see bootdev cmdline argument) might not be the first interface to come up, second if we are in guess-the-interface mode, multiple interface might receive different data for resolv.conf and we have to use the correct stuff to mount.
Quoted 3 lines
Quoted 1 line
So please revert that commit.
Do you agree the one copied ultimately should be the interface that successfully netbooted?
Well if you mean by that the one ultimately copied to the booted system that is correct. Just use the one /etc/resolv.conf after a successful mount.
Regards,
Philippe
Quoted 2 lines
Warren
--
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 14 lines · Seewer Philippe
Quoted 11 lines · Warren Togami
Quoted 10 lines · Seewer Philippe
Having only one resolv.conf is ok for a booted client. That is the normal and the usual behaviour is that the first interface to be up and ready with data is the one to write the contents.
But inside dracut we really do not want that behaviour. First the primary interface (see bootdev cmdline argument) might not be the first interface to come up, second if we are in guess-the-interface mode, multiple interface might receive different data for resolv.conf and we have to use the correct stuff to mount.
Quoted 9 lines
Quoted 4 lines
Quoted 1 line
So please revert that commit.
Do you agree the one copied ultimately should be the interface that successfully netbooted?
Well if you mean by that the one ultimately copied to the booted system that is correct. Just use the one
/etc/resolv.confafter a successful mount.
modules.d/40network/netroot:
[ -e /tmp/net.$netif.gw ] && . /tmp/net.$netif.gw
[ -e /tmp/net.$netif.hostname ] && . /tmp/net.$netif.hostname
[ -e /tmp/resolv.conf ] && cp -f /tmp/resolv.conf /etc/resolv.conf
Previous to this change:
[ -e /tmp/net.$netif.gw ] && . /tmp/net.$netif.gw
[ -e /tmp/net.$netif.hostname ] && . /tmp/net.$netif.hostname
[ -e /tmp/net.$netif.resolv.conf ] && cp -f /tmp/net.$netif.resolv.conf /etc/resolv.conf
Even the previous version appears to not guarantee that /etc/resolv.conf is the interface that successfully mounted the rootfs because this happens before the mount handler in the netroot script.
Could you please suggest a patch? Simply reverting it doesn't seem to improve things.
Warren Togami
[email protected]
--
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