Specifically document how to handle the case of explicitly starting a
bridge and then dynamically adding interfaces to it.
brctl_br0=''
bridge_add_eth0='br0'
This can be used for multi-homed connections and other advanced routing
in Linux. See the documentation links for more information about doing
this in linux.
The code was a originally pure addon into the conf.d/net files, written
in mid-2004 for doing multi-homing between two internet connections. I
have finally cleaned this up and integrated it. Thanks to Jonathan Kwan
for giving me the original impetus to develop this for Gentoo (it was
his dual internet connections...).
In the intervening years, it was a example of postup/postdown in the
net.example file, however that suffered from a few corner case issues.
If you were using the code from net.example, please see the updated
section 'Advanced Routing' on syntax, and drop your old function blocks.
Additionally, note that the rules added are now directly saved for
removal when the interface is taken down.
- Use sysfs to read bridge information from the system instead of
parsing the brctl outputs.
- Allow setting of all bridge configuration parameters using new sysfs
methods, modelled after bonding configuration. Also works for per-port
bridge interface parameters.
- Document pre-starting an empty bridge for dynamic add.
- Check for interface existence before adding to bridge.
- Should fix bug #293046, #309185.
The _netdev option in fstab on linux systems indicates
that a filesystem should only be mounted if the network is available.
This commit fixes netmount to support this.
Under normal conditions, runscript creates one child and waits for its
termination, which is signaled by a pipe write from the SIGCHLD
sighandler.
When running killprocs however more than one SIGHCLD signal is generated, at
least on all of my amd64 boxes running on real hardware and in vmware.
When the first SIGCHLD occurs svc_exec leaves the loop and closes the pipe.
Subsequent SIGCHLDs during the close can lead to a race condition and create an
EBADF error in the pipe write (pipe is closed but the file handle is still !=
-1).
We avoid this by blocking SIGHCHLD during the pipe close.
This reverts commit 4ea75dd1d6.
This caused a regression, see bug #345795.
Ifplugd is only designed to support wireless interfaces that use the
older wireless extentions.
We need to warn users that they should be using /etc/conf.d/local
instead of /etc/conf.d/local.start and /etc/conf.d/local.stop. This adds
those warnings.
