| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
* Added more debug_log calls around taking down interfaces.
* When taking down a bridge, put the member interfaces down too.
* Configure wireless interfaces before everything else - it might be that
a wireless interface is part of a bond/bridge/VLAN.
* When taking down VLAN interfaces, check if the underlying interface has
an IP address. If it does, leave it up; else take it down.
|
|
|
|
| |
Also removed redundant setting of a 0.0.0.0 IP address.
|
|
|
|
|
|
|
|
|
|
| |
Previously we had an xOPTS variable for each type of interface that could
be configured. I've replaced this with a generic IFOPTS option which can
be used to supply interface type specific options to 'ip'.
As the option is now generic, it can be applied to physical interfaces too;
so interfaces such as eth0 can have options applied just like a virtual
interface.
|
| |
|
|
|
|
|
|
|
| |
Before, it was not possible to set bridge specific options when creating
a bridge. This option adds the ability to set options via a BROPTS
variable, which takes a pipe (|) delimited set of options as documented
in the ip-link(8) man page (search for "BRIDGE Type Support").
|
| |
|
|
|
|
|
| |
This variable can be used to set custom options to configure the VLAN
interface. It does not need to be used in 'normal' operation with a VLAN.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The bonding code introduces three new variables for use in rc.inet1.conf:
BONDNICS[x]=""
BONDMODE[x]=""
BONDOPTS[x]=""
The BONDNICS variable takes a list of interfaces which should be slaved to
the bond.
BONDMODE sets the mode of the bond. Useful options are 'balance-rr',
'active-backup' and '802.3ad'. A full list of options can be found in
/usr/src/linux/Documentation/networking/bonding.txt and README.bonding.
BONDOPTS are a pipe (|) separated list of options to apply to the bond after
the interfaces have been added. Useful options are 'primary' (which is
required when BONDMODE is 'active-backup', 'xmit_hash_policy' (which can be
used with 'balance-rr' to choose the hash policy, 'lacp_rate' (which should
be used with '802.3ad' mode), and 'miimon' (which should be used with all
modes) See the kernel source documentation or README.bonding for more info.
|
|
|
|
|
|
|
|
|
|
|
| |
Remove code to skip interfaces that do not have IP addresses configured.
This checking is redundant now, and prevents interfaces being brought up
without IP addresses - which is important for bridging/bonding/VLANs.
When grabbing the vlan ID from the interface name, use ## instead of #.
Move the point where bridges are taken down to before the point where
wireless interfaces are taken down. This is required so that a bridge
on a wireless interface is take down before the interface itself.
When looping through the interfaces to take down, use MAXNICS-1.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
rc.inet1:
* VLAN support added to rc.inet1{,.conf}.
* Created new info_log function to replace $LOGGING variable.
* When logging to syslog, use a tag of 'rc.inet1' with PID.
* Decrease script noisyness at non-debug levels.
* Add more debugging logging for testing.
* Use -4 when configuring IPv4 loopback address.
* Configure ::1 for IPv6 loopbacking.
* Use 'local' not 'declare' in functions.
* Move where DAD is disabled when setting IPv6 addresses.
* Add error checking to some calls to 'ip'.
* Walk backwards through interfaces when deconfiguring.
rc.wireless:
* Log to syslog, failling back to stdout.
* Output error to stderr if not called from rc.inet1.
Manpages and netconfig:
* Updated for new VLAN support.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
It occurred to me that IPv6 addresses don't really need to have IP6ALIASES
since all IPs are just added to the interface itself. So I've changed the
syntax from using IP6ADDR + PREFIXLEN + IP6ALIASES to using just a single
IP6ADDRS array, which takes a list of IP addresses with a prefix length
in a space separated list. eg:
IP6ADDRS[0]="a:b:c:d::1/64 1:2:3::1/48"
If the prefix length is omitted a /64 length is assumed and a warning
emitted.
|
| |
|
|
|
|
|
|
| |
This is ugly, but it beats doing the sleep regardless of whether
there's any need for it and causing folks to whine about two
seconds of loss in boot time.
|
|
|
|
| |
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
|
|
| |
Thanks to David Spencer for much of this.
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
|
|
|
| |
This prevents an annoying "RTNETLINK answers: File exists" warning
when restarting networking
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
| |
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
| |
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
| |
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
| |
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
|
|
|
|
|
|
| |
This has the effect of allowing someone to disable ipv6 entirely,
and the script do the right thing.
Move the enabling of DAD into the correct section - no idea how
it ended up in the ipv4 block of code. D'oh.
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* USE_AUTOCONF->USE_SLAAC to be more consistent with IPv6 terminology
* PREFIX6->PREFIXLEN to be more consistent with IPv6 terminology
Thanks to David Spencer (idlemoor) for the patch to the above
* AUTOCONF_TIMEOUT->SLAAC_TIMEOUT to be consistent with USE_SLAAC change
Since some options for IPv6 no longer end in '6', we can relax the naming
of others, so these make sense: IPADDR6->IP6ADDR ; IPALIASES6->IP6ALIASES
Update to rc.inet1.conf example file - thanks to David Spencer.
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
|
|
|
|
|
| |
Also fix loading of interface modules - previously just $i was used,
where it should have been ${i%%:*}
Also added pre-loading of ipv6 module
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
|
|
|
|
|
|
| |
https://www.agwa.name/blog/post/beware_the_ipv6_dad_race_condition
This removes the need to sleep for 3 seconds after setting up the
interfaces, but does mean we lose detection of address clashes on
the inteface - it's a trade off
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
|
|
|
|
|
|
|
| |
In testing here, this was necessary to prevent daemons reporting 'Cannot
assign requested address' for IPv6 addresses when they attempt to bind()
to the interface address. This only seems to affect v6 addresses - v4
addresses seem to be immediately available to bind() - but a 3 second
pause shouldn't hurt given that the other methods of configuration can
block for up to 15 seconds.
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
| |
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
| |
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
|
|
| |
Removed unneeded sleep when taking interface down
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
|
|
| |
This is more in line with Slackware tradition
Signed-off-by: Robby Workman <rworkman@slackware.com>
|
|
|
|
| |
Signed-off-by: Robby Workman <rworkman@slackware.com>
|