Age | Commit message (Collapse) | Author |
|
Bug 4977
Need to wait to set/check bond primary device until after it
is put in bond group.
|
|
bootup
|
|
* fix this for eth, eth.vif, bond, bond.vif interfaces
* also, bridge-group code for bonding interfaces wasn't changed
with commit ec080f99 that re-strucutred bridging code in eth and
eth.vifs to fix bug 4708. bridge-group code for bond and bond.vif
interfaces is now similar to the fix for 4708
|
|
Bug 5256
Going out to perl to repeatedly validate ethernet device name slows
down adding VLAN's and other operations with lots of sub-devices.
|
|
|
|
The overhead of compiling perl script repeatedly slows down VIF creation
|
|
We need to delay disabling flow control in the sequence of events
until after the interface address has been set. An additional benefit
of adding a "priority" field is that interface configuration will not
fail if the attempt to disable flow control fails for some reason.
|
|
Want to process loopback first, so any local addresses assigned
to loopback are done.
|
|
All the other interface types are in vyatta-cfg-system, only
loopback and ethernet were in vyatta-cfg
|
|
|
|
got "invalid variable reference (invalid format)"
- need extra escape magic to call vyatta-cli-expand-var.pl.
|
|
|
|
|
|
|
|
|
|
Other half of Bug 5014
|
|
|
|
|
|
|
|
Bug 4918
|
|
Since kernel driver does not handle mode changes properly. Workaround
by removing slaves (and readding) before changing mode.
Bug 4758
|
|
The last change broke bond-group setup.
|
|
Use similar action to new bridge-group
Bug 4768
|
|
Use similar action to new bridge-group
Bug 4768
|
|
Unless miimon parameter is set, bonding device won't watch for
link changes (which makes it useless for any backup modes).
|
|
Bug 4672
Use ip command rather than sysfs to avoid quoting issues
(cherry picked from commit 1b4902564b5a853b925d2c239c8e6a39b1734433)
|
|
bug 4694
No reason to block disabling bonding interface with VIF
Kernel handles it fine.
(cherry picked from commit 3c76bd6b12dd211075d1dd689a2772c534ec7abe)
|
|
Workaround for bonding VLAN_CHALLENGED bug 4694
(cherry picked from commit ca8bff3d63074c52cf9b532eeec3a6a4cfe76722)
|
|
(cherry picked from commit c43e9e86ea502c455dcffac10d50304d47cad8ba)
|
|
(bug 4656)
(cherry picked from commit a054f90f492653c34d8c1c6443b1b132bc9fbc22)
|
|
(cherry picked from commit ed127b43756dfe852cff453ac8d8a72c01cbefd2)
|
|
ifenslave is the old way, and may not be supported in future.
Use sysfs instead to add/delete slaves.
(cherry picked from commit 0bd1cd12ccb047afce7444f5728fc3bbc3f328a5)
|
|
Macvlan and vlan tagging doesn't work right in the kernel (2.6.29)
so wait till it is fixed upstream.
(cherry picked from commit aaed494ef8bbe580da4d378e4844bb0203987d97)
|
|
These are all templates that do nothing in current config
and only contain a useless help: value.
(cherry picked from commit d932a2d53618d38a3a5d4c668e18fc5920e0ad6b)
|
|
Bug 4513
(cherry picked from commit 037c5999e70345243ca1a4500a5e247109f54830)
|
|
Bug 4511
Since loading bond module is broken in 2.6.29 or later kernel
and the root cause is in sysfs/procfs which is a mess.
Workaround the problem by loading bond module once and using sysfs.
(cherry picked from commit 47f054cb196959bb79960f3b6c1524b4ff7d32a8)
Conflicts:
templates/interfaces/bonding/node.def
|
|
Update makes no sense on a multi-valued node, only create or delete
have meaning.
(cherry picked from commit 273c1418553a39dcb4417acbf70cb863c97d1ccd)
|
|
This change cleans up the bridging configuration, uses the new COMMIT_ACTION
environment variable, and avoids using state files to pass information between
templates.
|
|
|
|
interfaces address.
|
|
Bug 4478
Incorrect syntax on ip link command.
|
|
|
|
|
|
|
|
Avoid name-conflict with Xen
|
|
|
|
Some interfaces have link-detect others do not, and vyatta-cfg-quagga
is trying to handle generic case.
|
|
Change to use script for allowed values.
|
|
If using STP then apply configured forwarding-delay or 15 if none
specified.
|
|
Bug 4226
There is no need for forwarding delay unless doing spanning tree.
|