]> git.proxmox.com Git - mirror_ifupdown2.git/log
mirror_ifupdown2.git
9 years agoAllow customer set bond defaults for CL with ifupdown2
Sam Tannous [Wed, 22 Jul 2015 22:38:07 +0000 (18:38 -0400)]
Allow customer set bond defaults for CL with ifupdown2

Ticket: CM-6723
Reviewed By: roopa
Testing Done: unit tested and testifupdown2 test suite

This patch installs bond interface defaults in

     /etc/network/ifupdown2/policy.d/bond_defaults.json

and allows users to modify this file.   Users can then leave out these
bond attributes in their configs to save typing and space.

It also changes the ifenslave and ifenslaveutil module to bond and
bondutil, respectively to be consistent with other modules
(and also because customers think of "bond" interfaces not
"ifenslave" interfaces.)

For example, the default file installed looks like the following:

{
    "README": "This file is user generated and modifiable.",
    "bond": {
        "defaults": {
                "bond-mode": "802.3ad",
                "bond-miimon": "100",
                "bond-use-carrier": "1",
                "bond-lacp-rate": "0",
                "bond-min-links": "1",
                "bond-xmic-hash-policy": "layer3+4"
         }
    }
}
Please enter the commit message for your changes. Lines starting

9 years agoMerge branch 'CumulusLinux-2.5_br' of ssh://stash.cumulusnetworks.com:7999/ccmr/cumul...
Roopa Prabhu [Mon, 13 Jul 2015 22:02:14 +0000 (15:02 -0700)]
Merge branch 'CumulusLinux-2.5_br' of ssh://stash.cumulusnetworks.com:7999/ccmr/cumulus into CumulusLinux-2.5_br

9 years agoFix ip addr parsing breakage on sid
Roopa Prabhu [Mon, 13 Jul 2015 21:33:01 +0000 (14:33 -0700)]
Fix ip addr parsing breakage on sid

Ticket:
Reviewed By: stannous
Testing Done: sanity testing

9 years agoapt-get remove of python-ifupdown2 fails due to postrm script error
Sam Tannous [Mon, 13 Jul 2015 15:12:44 +0000 (11:12 -0400)]
apt-get remove of python-ifupdown2 fails due to postrm script error

Ticket: CM-6655
Reviewed By: Trivial
Testing Done: tested install and removal of python-ifupdown2

There was an invalid equality test in the postrm script for
ifupdown2.  In Dash, equality is tested with a single '='
and not '=='.

9 years agoMove interface name check to networkinterfaces.py where all other parse
Roopa Prabhu [Thu, 9 Jul 2015 18:07:06 +0000 (11:07 -0700)]
Move interface name check to networkinterfaces.py where all other parse
errors are reported

Ticket: CM-5882
Reviewed By: stannous
Testing Done: Tested with a long interface name

9 years agoifupdown2 ethtool add-on module should not get unnecessary running configs for perfor...
Sam Tannous [Thu, 9 Jul 2015 17:40:00 +0000 (13:40 -0400)]
ifupdown2 ethtool add-on module should not get unnecessary running configs for performance reasons

Ticket: CM-6630
Reviewed By: roopa
Testing Done: unit tested

We still run ethtool on most interfaces even if we don't have defaults to use.
The last bug I fixed (CM-6557) should have also fixed that (even if it was specific to
ifquery).

We now check if there is a configured attribute and if there is a default
value. If we do not have both, we continue without calling ethtool to get the running value.

9 years agovxfld: ifupdown2 addon for vxrd
Sameer Ladiwala [Mon, 6 Jul 2015 20:40:52 +0000 (13:40 -0700)]
vxfld: ifupdown2 addon for vxrd

Ticket: CM-6358
Reviewed By: roopa
Testing Done: Yes

ifupdown2 creates overrides vxrd's config when either (or both) the vxrd-src-ip and vxrd-svcnode-ip appear under the loopback interface's configuration.
Example
address 12.0.0.4/32
vxrd-src-ip 12.0.0.4
vxrd-svcnode-ip 99.99.99.9

9 years agoRunning ifquery --running -a will truncate the pvid number to a single digit
Sam Tannous [Thu, 2 Jul 2015 21:14:24 +0000 (17:14 -0400)]
Running ifquery --running -a will truncate the pvid number to a single digit

Ticket: CM-6284
Reviewed By: roopa
Testing Done: unit tested change

Running ifquery --running -a will truncate the pvid number to a single digit for the bridge.
This patch appendis the entire pvid instead of only the first character.

9 years agoMake ifupdown2 print meaningful error when interface name length exceeds 15 characters
Sam Tannous [Thu, 2 Jul 2015 21:10:04 +0000 (17:10 -0400)]
Make ifupdown2 print meaningful error when interface name length exceeds 15 characters

Ticket: CM-5882
Reviewed By: gospo
Testing Done: unit tested

When a user enters an interface name longer than 15 characters,
the error message is not clear about what the problem is.

   warning: netlink: Numerical result out of range <<<<<<<<<<<<<<<<

This patch catches the error before netlink gets the call and
prints

    error: the following interface names are too long: bond-xconnect.1006

9 years agoifquery --running should not show link attributes on logical interfaces
Sam Tannous [Thu, 2 Jul 2015 19:59:34 +0000 (15:59 -0400)]
ifquery --running should not show link attributes on logical interfaces

Ticket: CM-6557
Reviewed By: roopa, gospo
Testing Done: unit tested

ifquery should not show link attributes for interfaces we do not have defaults for.
Although somewhat useful, we don't want people saving the bond link attributes
to try to use then in a config

9 years agoFix warning (was emitting two lines)
Roopa Prabhu [Wed, 1 Jul 2015 22:01:15 +0000 (15:01 -0700)]
Fix warning (was emitting two lines)

Ticket:
Reviewed By: trivial
Testing Done:

9 years agoun-glob expression in sample file
Roopa Prabhu [Wed, 1 Jul 2015 22:00:07 +0000 (15:00 -0700)]
un-glob expression in sample file

Ticket:
Reviewed By: trivial
Testing Done: Tested that sample works fine

9 years agoifreload causes unnecessary dhcp restart for eth0
Sam Tannous [Tue, 30 Jun 2015 17:47:08 +0000 (13:47 -0400)]
ifreload causes unnecessary dhcp restart for eth0

Ticket: CM-5730
Reviewed By: roopa
Testing Done: tested change on Sean and Jason's lab setup

The dhcp restart on all interfaces using dhcp causes a reconnection
for NTP and in addition, a recent addition of rack management packages
(CM-5782) added ntp and a exit-hook script that also interrupted NTP
for a second time.   This patch will check for dhclient running (IPv4 and IPv6)
and not restart dhcp if it's already running.

9 years agoFix check for bridge interface in scheduler during upperiface check
Roopa Prabhu [Sat, 27 Jun 2015 17:12:21 +0000 (10:12 -0700)]
Fix check for bridge interface in scheduler during upperiface check

Ticket: CM-6525
Reviewed By: trivial
Testing Done: Tested with failing config mentioned in the bug

c4d27f53e965af5edc938acae25bcb249a2214a7 introduced a
new interface kind of BRIDGE_VLAN_AWARE which broke the
scheduler check. This patch fixes the check.

9 years agoifupdown2 should not allow IP address command under vlan aware bridge stanza
Sam Tannous [Wed, 24 Jun 2015 15:39:56 +0000 (11:39 -0400)]
ifupdown2 should not allow IP address command under vlan aware bridge stanza

Ticket: CM-6106
Reviewed By: roopa,scotte
Testing Done: ran ifupdown2 regression suite and hand tested

This patch prevents IP address assignment under vlan aware bridges.
This prevents some confusion because a bridge subinterface is required
for layer 3 termination.

9 years agoskip adding filtered or blacklisted interfaces in the dependency graph
Roopa Prabhu [Mon, 22 Jun 2015 00:37:27 +0000 (17:37 -0700)]
skip adding filtered or blacklisted interfaces in the dependency graph

Ticket:
Reviewed By: CCR-3043
Testing Done:

This was resulting in lower interfaces not being brought up at boot
when their upper-ifaces were not-marked auto. And this was because
the dependency graph continued to have the non-auto upper interfaces
and the whole tree was ignored by the scheduler.

This fix eliminates filtered interfaces from the dependency graph
to simplify the scheduler and also to avoid scheduler picking up
non-auto (or non-class interfaces).

testcase (Without this patch swp1 was skipped at boot)
iface swp1.100

auto swp1
iface swp1

9 years agoEliminate confusing and unnecessary ethtool calls
Sam Tannous [Sat, 20 Jun 2015 01:27:49 +0000 (21:27 -0400)]
Eliminate confusing and unnecessary ethtool calls

Ticket: CM-6275
Reviewed By: roopa,dustin
Testing Done: Unit tested and ran complete ifupdown2 test suite

There are unnecessary calls to the ethtool module that cause some error messages to be printed.
We currently don't check ifaceobj.link_kind at the start
so this check was added.

If someone creates a new interface without setting link_kind, we still printed error message.
This patch adds a check at the start of ethtool.

This patch also removes the global link defaults (used by ifupdown2/ethtool)
created when update-ports is called.  The defaults are created for each swp port
so we don't have defaults (and therefore will not attempt to set) for interfaces
like eth0, loopbacks, dummys, vxlans, or any other non swp interface.

9 years agoifupdown changes for active-active vxlan
Balakrishnan Raman [Thu, 18 Jun 2015 08:18:43 +0000 (01:18 -0700)]
ifupdown changes for active-active vxlan

Ticket: CM-3715
Reviewed By: CCR-3065
Testing Done: unit testing

Changes to proto down vxlan devices to avoid packet loops or duplicates before
clagd checks consistency and installs egress acl rules. When vxlan device is
brought up, check for presence of peer link and proto-down the device if peer
link exists.
- get_dependent_ifacenames is used to eliminate order dependency in specifying
peer link and vxlan configuration in /etc/network/interfaces.
- A separate script clagVxlanProtoDown is used to protodown vxlan devices if
peer link is configured separately (ifup peerlink) and vxlan devices are
already functional in the system.

9 years agoEnhance globbing in ifupdown2 to support breakout cables
Sam Tannous [Wed, 17 Jun 2015 00:58:15 +0000 (20:58 -0400)]
Enhance globbing in ifupdown2 to support breakout cables

Ticket: CM-5365
Reviewed By: roopa,dwalton
Testing Done: Ran regression suite and hand tested.

This patch extends the globbing support in ifupdown2
to handle two levels of square brackets so that breakout
cables can be handled (e.g. swp[2-7]s[0-3]).  Also modified
the user's guide and the man pages.

9 years agovxlan error 'str' object has no attribute 'append'
Sam Tannous [Tue, 16 Jun 2015 21:15:57 +0000 (17:15 -0400)]
vxlan error 'str' object has no attribute 'append'

Ticket: CM-6329
Reviewed By: roopa
Testing Done: unit testing with vxlan configs

A bug iproute2.py creates an unnecessary warning
about appending the ageing value to a string.
This patch removes the append and simply sets the ageing value.

9 years agoDocument exclude (-X) option in ifreload
Sam Tannous [Wed, 20 May 2015 14:15:39 +0000 (10:15 -0400)]
Document exclude (-X) option in ifreload

Ticket: CM-5419
Reviewed By: Trivial
Testing Done: built ifupdown2 deb and tested on switch

This patch adds man page mention of undocumented (but working)
option to exclude interfaces from a reload (-X option)
(cherry picked from commit 0358905782a4c0290a967d6f730fdcb27f2af3d9)

9 years agoDon't allow IP addresses on ports enslaved in bonds or bridges
Sam Tannous [Wed, 3 Jun 2015 00:35:08 +0000 (20:35 -0400)]
Don't allow IP addresses on ports enslaved in bonds or bridges

Ticket: CM-5146
Reviewed By: roopa,jtoppins
Testing Done: built new ifupdown package and ran testifupdown2 suite of tests

This patch prevents enslaved interfaces from having IP addresses.
(cherry picked from commit 0c00606fbc76db11557a8e946310e93a2b376aa7)
(cherry picked from commit dc30987acfc6af356b9e055db95d94ae45f0de9f)

9 years agoAdd default link parameter support for ethtool module
Sam Tannous [Sun, 10 May 2015 19:42:47 +0000 (15:42 -0400)]
Add default link parameter support for ethtool module

Ticket: CM-5254
Reviewed By: roopa
Testing Done: tested master and 2.5_br images with testifupdown2 suite and hand tested

This patch creates a json defaults file upon bootup
(which can be overridden by customer configs in /etc)
which the ethtool module in ifupdown2 will consult
when "link-x" configs are removed in order to restore
them to the initial settings used by the switch.
(cherry picked from commit 8388664f5a5a85f2a813cafbf40ac92d7b86f4bf)

Conflicts:
packages/cl-utilities/usrlib/update-ports
(cherry picked from commit 21c9c10ab2fccaf60be9accb337e82541d497cc4)

9 years agoAdded ifupdown2 support for vxlan-ageing config
Sam Tannous [Thu, 5 Mar 2015 17:54:55 +0000 (12:54 -0500)]
Added ifupdown2 support for vxlan-ageing config

Ticket: CM-5105
Reviewed By: roopa
Testing Done: tested configs before and after change

We needed a configuration option to add vxlan-ageing to
ifupdown2 configs.  This patch adds the option to change
the vxlan-ageing timer currently set with "ip link"
commands.
(cherry picked from commit 9832462c365bd2b900b98f5675d407d1b11c4a95)

Conflicts:
packages/ifupdown2/addons/vxlan.py
packages/ifupdown2/ifupdownaddons/iproute2.py

9 years agoAdded lacp bypass all-active parameter in ifupdown2 man page
Wilson Kok [Sat, 16 May 2015 18:06:10 +0000 (11:06 -0700)]
Added lacp bypass all-active parameter in ifupdown2 man page

Ticket: CM-6038
Reviewed By:
Testing Done:

9 years agoMove stale slave delete to after the adds are done
Roopa Prabhu [Fri, 15 May 2015 06:00:57 +0000 (23:00 -0700)]
Move stale slave delete to after the adds are done

Ticket: CM-5887
Reviewed By: trivial
Testing Done: Tested with sample config in the bug
(cherry picked from commit e90149a1e8ea3bc94e07fb080361e6909b12e81b)

9 years agoAdd new SKIP_DOWN_AT_SYSRESET /etc/init.d/networking config option to
Roopa Prabhu [Tue, 12 May 2015 14:32:53 +0000 (07:32 -0700)]
Add new SKIP_DOWN_AT_SYSRESET /etc/init.d/networking config option to
skip deconfiguring interfaces

Ticket: CM-5900
Reviewed By: CCR-2921
Testing Done: Tested 'service networking stop' during system reboot and shutdown

9 years agoLACP bypass fixes
Wilson Kok [Fri, 8 May 2015 18:17:14 +0000 (11:17 -0700)]
LACP bypass fixes

Ticket: CM-5924, CM-5937, CM-5921, CM-5932
Reviewed By: CCR-2911
Testing Done:

bonding driver:
- renamed bond lacp-bypass-use-priority config option to
  lacp-bypass-all-active with the default value of 0
- ignore clag role when lacp-bypass-all-active is enabled
- added info logging of lacp bypass state changes

iproute2:
- renamed bond lacp-bypass-use-priority config option

ifupdown:
- renamed bond lacp-bypass-use-priority config option
- added default value handling for lacp bypass attributes

switchd:
- macs learned on individual slaves are now sync to kernel with
  the bond's ifindex.  This eliminates flooding when cpu is
  communicating to the host where the bond is in lacp bypass
- fixed bond forwarding state propagation to individual slaves in
  bypass

9 years agomove udev script processing to before restarting networking
Roopa Prabhu [Thu, 7 May 2015 23:13:52 +0000 (16:13 -0700)]
move udev script processing to before restarting networking

Ticket: CM-5764
Reviewed By: jtoppins
Testing Done:

9 years agoAdded ifupdown support for different lacp bypass modes
Wilson Kok [Sat, 2 May 2015 22:19:51 +0000 (15:19 -0700)]
Added ifupdown support for different lacp bypass modes

Ticket: CM-5716
Reviewed By:
Testing Done:

Added mode to distinguish between priority based and multiple links
lacp bypass modes.

9 years agoMerge branch 'CumulusLinux-2.5_br' of ssh://dev.cumulusnetworks.com/home/trac/cumulus...
Roopa Prabhu [Thu, 30 Apr 2015 20:03:07 +0000 (13:03 -0700)]
Merge branch 'CumulusLinux-2.5_br' of ssh://dev.cumulusnetworks.com/home/trac/cumulus into CumulusLinux-2.5_br

9 years agoFix bridge pvid add/del order
Roopa Prabhu [Thu, 30 Apr 2015 19:59:16 +0000 (12:59 -0700)]
Fix bridge pvid add/del order

Ticket: CM-5832
Reviewed By: CCR-2875
Testing Done: Tested with test case in the bug and other orders

kernel(upstream and us) does not seem to honor the vidinfo flags
during the deletes. Hence, this reordering becomes necessary.

tested with the example in the bug and also some other examples.

code wise this combines two existing methods but leaves the two existing methods
around for future use. Will remove them if they become completely
unnecessary.

9 years agoFix ifreload man page to include the new behaviour where it does not
Roopa Prabhu [Tue, 28 Apr 2015 12:38:15 +0000 (05:38 -0700)]
Fix ifreload man page to include the new behaviour where it does not
down all changed interfaces

Ticket: CM-5819
Reviewed By: trivial
Testing Done: Tested man page
(cherry picked from commit 93ecf73b22d317dcf2a0924030f34f008fc25c33)

9 years agoDont down an interface during ifreload just because the number of iface
Roopa Prabhu [Tue, 28 Apr 2015 04:18:14 +0000 (21:18 -0700)]
Dont down an interface during ifreload just because the number of iface
sections for that interface changed

Ticket: CM-5841
Reviewed By:
Testing Done: tested ifreload with new iface section

<email_notes>
problem stmt:
if the user adds a new section for an existing interface, I mark the
interface for down.
Basically the below:

               if len(newifaceobjlist) != len(lastifaceobjlist):
                    ifacedownlist.append(ifname)
                    continue

You rarely need to add a new section to the interfaces file. It is not
recommended even by the user guide.
sankaran was trying to add a new address. Which he could have added to
the same iface section.

But, looking at his trivial example, i am thinking of not marking an
interface for down if the user merely tried to add a new section to an
existing interface
</email_notes>

9 years agoFix addons man page for a few mstpctl defaults
Roopa Prabhu [Mon, 27 Apr 2015 22:38:07 +0000 (15:38 -0700)]
Fix addons man page for a few mstpctl defaults

Ticket: CM-5813
Reviewed By: sashok
Testing Done: build and tested man page

9 years agoifupdown2: fixup maintainer scripts
Jonathan Toppins [Tue, 21 Apr 2015 21:07:46 +0000 (14:07 -0700)]
ifupdown2: fixup maintainer scripts

Ticket: CM-5764
Reviewed By: TBA
Testing Done: of course none ;) Will update when I have some time to test

fupdown2's postinst and postrm scripts do not conform to Debian's policy
for maintainer scripts [1]. Specific non-conformance follows:

* fixed maintainer scripts so they only use /bin/sh as other shells are
  non-essential packages and may not exist on the system
* fixed maintainer scripts so only actions specified by $1 are performed
  per run of the script
* fixed maintainer scripts to only support specific values for $1, and
  will otherwise exit with error
* included at the bottom of the postrm script the "#DEBHELPER#" tag as
  during package building debianhelper will inject extra code at this
  location

[1] https://www.debian.org/doc/debian-policy/ch-maintainerscripts.html
(cherry picked from commit 2289fe4e90a0134fcb07fe7e2e838de755d73e68)

9 years agoadd new ifupdown2.conf option ifreload_down_changed to control ifreload
Roopa Prabhu [Sat, 25 Apr 2015 22:33:28 +0000 (15:33 -0700)]
add new ifupdown2.conf option ifreload_down_changed to control ifreload
ifdown behaviour.

Ticket: CM-5819
Reviewed By: CCR-2846
Testing Done: tested ifreload evo test case

ifreload_down_changed is 0 by default which will make
sure ifreload will not execute down on changed interfaces
but only on deleted interfaces making it non-disruptive.

some notes from CCR:

ifreload was designed to be an optimization for 'service networking
restart' or 'ifdown -a + ifup -a'.
essentially it is a combination of 'ifdown + ifup' with some smarts in
which interfaces it will execute ifdown on.

By default it does the below:
ifdown all interfaces that were deleted from the interfaces file
ifdown all interfaces that were changed from the last time they were
ifup'ed
ifup -a (execute ifup on all interfaces marked auto)

Did not realize people will use ifreload as much as they do today. Also,
they may execute it on a production box when changes are made. ifdown on a production box can be
disruptive because if the ifdown which is part of the ifreload.

To have a non-disruptive option to ifreload, 2.5 added a new option -c
that only executed 'ifup' on all interfaces. Thus reloading all auto +
any other interfaces that were once brought up on the box (essentially
all interfaces present in the saved state file). This by default did not
do anything to the interfaces that got deleted from the file. But had an
ifupdown2.conf toggle to do so.

Looking at the evo use case, they do want to use a single command that
modifies, adds, deletes with
minimum disruption. we can achieve maybe what they want with multiple
commands (But there is also a case of a bug in the build evo is running
which makes it not so easy ).

This patch fixes the bug and also tries to change the default ifreload
behaviour controllable via a variable in ifupdown2.conf.
when ifreload_down_changed=0 in ifupdown2.conf, ifreload will only
ifdown interfaces that were deleted
from the file but not the ones that changed. subsequent ifup as part of
ifreload on the interfaces
that changed will apply the delta. And ifreload_down_changed default
value is '0'.

WIth the patch, ifreload by default will do the below (going back to the
previous default is just a toggle in the ifupdown.conf file):
ifdown all interfaces that were deleted from the interfaces file
ifup -a (execute ifup on all interfaces marked auto)

It sounds like a big change of behaviour for a hotfix release, but
essentially the patch just moves a few things around. And the change in
behaviour is so subtle that it is not very visible.
It just makes it non-disruptive.
(cherry picked from commit 2f7977834d4912a69159d27e54ba201f58a321d8)

9 years agoifupdown2 loses interfaces on second down of swp port
Sam Tannous [Fri, 24 Apr 2015 00:19:22 +0000 (20:19 -0400)]
ifupdown2 loses interfaces on second down of swp port

Ticket: CM-5693
Reviewed By: roopa
Testing Done: tested bridge and bonds with interfaces with configs

Both bridge and mstpctl modules set priv_flags on interfaces
that have configs (like link-speed) even when used as bridge-ports.
And this collision causes statemanager.ifaceobj_sync() to never get called
because ifaceobj.priv_flags is 1 (we return immediately):
The fix was to create a new iface module_flags array to carry module info.

9 years agoifupdown2: fixup maintainer scripts to not crash on files not existing
Jonathan Toppins [Mon, 20 Apr 2015 20:41:05 +0000 (13:41 -0700)]
ifupdown2: fixup maintainer scripts to not crash on files not existing

Ticket: CM-5735
Reviewed By: no one
Testing Done: lots of repo upgrade testing by John B.

This partially reverts commit 8959e8e4ec93487228733b17ef2b220ed47c1260 ("Fix
bash completion for sudo users (mostly cosmetic)") in the sections where
the Debian maintainer scripts were modified.

In addition fixed up the Debian maintainer scripts to prevent a
potential upgrade failure, discussion follows:

The Cause For the upgrade Failure
=================================

Commit 8959e8e4ec93487228733b17ef2b220ed47c1260 introduced changes to
ifupdown2's package maintainer scripts to override some files provided
by the bridge-utils and udev packages as they conflict with ifupdown2.

The change to ifupdown2's postrm script is currently causing the script
to exit with a failure code if the following files do not exist on the
box:
/etc/udev/rules.d/80-networking.rules
/etc/udev/rules.d/60-bridge-network-interface.rules

The reason for this is in how the script is written.

Since the postrm script is configured to exit on the first command
failure, the "set -e" on line 3, the commands on lines 20 and 22 will
fail if the files being looked for do not exist, causing the script to
exit with a non-zero exit code. Causing apt to fail the upgrade.

.postrm script excerpt
----------
20 udevlink=$(readlink /etc/udev/rules.d/80-networking.rules 2>/dev/null)
21 [ -n "$udevlink" -a "$udevlink" == "/dev/null" ] && rm -f /etc/udev/rules.d/80-networking.rules
22 udevlink=$(readlink /etc/udev/rules.d/60-bridge-network-interface.rules 2>/dev/null)
23 [ -n "$udevlink" -a "$udevlink" == "/dev/null" ] && rm -f /etc/udev/rules.d/60-bridge-network-interface.rules
----------

Solution
--------

The solution requires having the newer version of the ifupdown2 package
provide a preinst script that gets run before the older package's postrm
script is run[1,2]. This gives us the opportunity to quietly fix the
problem. We also fix the postrm script in newer versions of the package
so it cannot bomb out in this way.

Finally, removed what appears to be a poor merge from commit
5f5d84e3261bf70dd3541666a3d3a7f817727ce6.

Fixes: 8959e8e4ec93487228733b17ef2b220ed47c1260
(cherry picked from commit 0eab79eb5aed4cdf0674e3f0ddad3631dd457bdc)

9 years agoModified ifupdown support for vxlan head end replication
Wilson Kok [Wed, 15 Apr 2015 16:48:12 +0000 (09:48 -0700)]
Modified ifupdown support for vxlan head end replication

Ticket: CM-5637
Reviewed By: CCR-2755
Testing Done:

Replaced vxlan-peernodeip attribute with vxlan-remoteip.  Updated
the vxlan and iproute2 modules to configure head end replication
using bridge fdb commands.

9 years agorework error message
Roopa Prabhu [Sat, 4 Apr 2015 05:33:53 +0000 (22:33 -0700)]
rework error message

Ticket: CM-5373
Reviewed By:
Testing Done: tested interfaces file with shared dependents

9 years agoAdd check for shared dependents during building dependency list
Roopa Prabhu [Sat, 4 Apr 2015 04:24:25 +0000 (21:24 -0700)]
Add check for shared dependents during building dependency list

Ticket: CM-5373
Reviewed By: sam
Testing Done: Tested with shared slaves in bridge and bonds

9 years agoDont follow dependents during down of interfaces as part of ifreload
Roopa Prabhu [Sat, 4 Apr 2015 03:47:06 +0000 (20:47 -0700)]
Dont follow dependents during down of interfaces as part of ifreload

Ticket: CM-5571
Reviewed By:
Testing Done: tested ifreload with example in the bug

ifdown of the bridge svi was following the lowerdev and downing the
bridge. kernel implicitly deletes all svi's on bridge down.
ifup of the bridge usually handles it, but a change done in 2.5.1
was causing it not to. Thinking about this some more, its safe
to not follow dependents on down during a ifreload. This patch does
just that.

9 years agoAdd config len check to iface compare
Roopa Prabhu [Sat, 4 Apr 2015 03:33:28 +0000 (20:33 -0700)]
Add config len check to iface compare

Ticket: CM-5253
Reviewed By: sam
Testing Done: Tested ifreload with steps in the bug

9 years agoPrevent MTU from being set to 1500 on interface downing.
Sam Tannous [Wed, 18 Mar 2015 01:59:15 +0000 (21:59 -0400)]
Prevent MTU from being set to 1500 on interface downing.

Ticket: CM-5355
Reviewed By: roopa
Testing Done: Tested config in ticket

MTU settings are defaulted 1500 when interface is ifdowned.
This has the effect of changing the MTU on the interface and any subinterfaces to
1500.  And if these subinterfaces are in a bridge, the bridge will pick the MIN MTU
so the bridge keeps this MTU even after the interface is brought back up. The sub
interface does not change to a higher value then 1500 (kernel VLAN driver only
moves the MTU is the decreasing direction.

9 years agoMerge branch 'CumulusLinux-2.5_br' of ssh://dev.cumulusnetworks.com/home/trac/cumulus...
Roopa Prabhu [Tue, 10 Mar 2015 21:19:41 +0000 (14:19 -0700)]
Merge branch 'CumulusLinux-2.5_br' of ssh://dev.cumulusnetworks.com/home/trac/cumulus into CumulusLinux-2.5_br

9 years agomerge docs.addons content into docs directory
Roopa Prabhu [Tue, 10 Mar 2015 21:18:30 +0000 (14:18 -0700)]
merge docs.addons content into docs directory

Ticket:
Reviewed By:
Testing Done:

9 years agoCorrect error message for unsupported ifupdown2 attributes
Sam Tannous [Tue, 10 Mar 2015 20:46:07 +0000 (16:46 -0400)]
Correct error message for unsupported ifupdown2 attributes

Ticket: CM-5349
Reviewed By: Trivial
Testing Done: Tested on cel box

Corrected a minor error in how we handle unsupported config attributes.
Instead of

     error: not enough arguments for format string

we will now print this:

     warning: peerlink.4000: unsupported attribute 'clagd-foobar'
(cherry picked from commit 44c6004a33d59ee234f8ee3d5f450e158c9e5bdc)

9 years agoWrong initialization of iface link kind (still puzzled how i had it this
Roopa Prabhu [Thu, 26 Feb 2015 22:26:58 +0000 (14:26 -0800)]
Wrong initialization of iface link kind (still puzzled how i had it this
way)

Ticket:
Reviewed By: trivial
Testing Done: Tested interface upperiface bring up

This was causing the auto upperiface bringup to pick all upperifaces on
malliks setup. This is a large number in a 500 svi setup.

9 years agomove link_set 'up' errors to debug
Roopa Prabhu [Wed, 25 Feb 2015 21:45:37 +0000 (13:45 -0800)]
move link_set 'up' errors to debug

Ticket: CM-5199
Reviewed By: trivial
Testing Done: tested with interfaces file in the bug

9 years agoIncrease LACP bypass period max to 900 seconds
Wilson Kok [Mon, 23 Feb 2015 19:18:34 +0000 (11:18 -0800)]
Increase LACP bypass period max to 900 seconds

Ticket: CM-5000
Reviewed By: Roopa
Testing Done:

9 years agoMerge branch 'CumulusLinux-2.5_br' of ssh://dev.cumulusnetworks.com/home/trac/cumulus...
Roopa Prabhu [Wed, 18 Feb 2015 20:32:05 +0000 (12:32 -0800)]
Merge branch 'CumulusLinux-2.5_br' of ssh://dev.cumulusnetworks.com/home/trac/cumulus into CumulusLinux-2.5_br

9 years agoCall mako renderer before rendering backslashes ('\')
Roopa Prabhu [Wed, 18 Feb 2015 20:28:50 +0000 (12:28 -0800)]
Call mako renderer before rendering backslashes ('\')

Ticket: CM-5104
Reviewed By:
Testing Done: Tested with interfaces file listed in the bug

9 years agoReplace bond-slave "link down" with "link protodown" for CLAG bond proto_down
anuradhak [Wed, 18 Feb 2015 16:03:07 +0000 (08:03 -0800)]
Replace bond-slave "link down" with "link protodown" for CLAG bond proto_down
handling.

Ticket: CM-4802
Reviewed By: CCR-2565
Testing Done: Yes

1. Earlier various clag components where using link admin state for enforcing
proto_down. With the introduction of a seperate proto_down state this is no
longer needed. Now on clag bond "proto_down" the bond and its slaves are
explicitly placed in a proto_down state. This carrier downs the slaves and
oper-downs the clag-bond. Change has been done in clagd, clag-monit script and
bonding driver.
2. Earlier misc tricks where needed in ifupdown2 to prevent slave state changes
for clag bond outside of clagd. These tricks are no longer needed and have been
reverted to keep ifupdown2 handling simple/consistent. This change basically
reverts #6985e1a376b538fbfa2346ef657427ae7e8f117a.

9 years agoifupdown2 must handle JSON input that is non-list
Sam Tannous [Wed, 11 Feb 2015 16:37:41 +0000 (11:37 -0500)]
ifupdown2 must handle JSON input that is non-list

Ticket: CM-4417
Reviewed By: roopa
Testing Done: Build powerpc image and tested alternate json format

ifupdown2 was patched to handle nonlist JSON input since this
is a valid format.
(cherry picked from commit 2597194f6f34344495f3a2b44bfe1d05887e1e77)

9 years agoAdd checks for BRIDGE_VLAN interfaces during link up/down
Roopa Prabhu [Tue, 10 Feb 2015 19:28:53 +0000 (11:28 -0800)]
Add checks for BRIDGE_VLAN interfaces during link up/down

Ticket: CM-4718
Reviewed By:
Testing Done: tested ifup/ifdown of vlan interfaces

9 years agoCheck for zero address on bridge ports before enslaving it to the bridge
Roopa Prabhu [Tue, 10 Feb 2015 19:17:00 +0000 (11:17 -0800)]
Check for zero address on bridge ports before enslaving it to the bridge

Ticket: CM-4859, CM-4382
Reviewed By:
Testing Done: Tested with config which included bond bridge ports with
zero address on the bond.

The bridge driver rejects the port add with -EINVAL.
A bond with no slaves is one usecase where the bond interface
has a zero hw address and is rejected by the bridge.
EINVAL currently is very confusing.

This patch checks for a valid ether add before enslaving the port to the
bridge.

9 years agoAdd dependency on python-argcomplete, python-ipaddr and python-gvgen
Roopa Prabhu [Mon, 9 Feb 2015 21:11:16 +0000 (13:11 -0800)]
Add dependency on python-argcomplete, python-ipaddr and python-gvgen

Ticket: CM-4453
Reviewed By:
Testing Done: Build tested
(cherry picked from commit ee9aa1a1bf510f950030f72cbc12ed41b0bc880d)

9 years agoMove clag_enable setting to ifupdown2 ifenslave.
anuradhak [Mon, 2 Feb 2015 17:54:53 +0000 (09:54 -0800)]
Move clag_enable setting to ifupdown2 ifenslave.

Ticket: CM-4802
Reviewed By:
Testing Done: yes

1. clag_enable needs to be set before slaves are added to the bond. Otherwise
bonding driver will bring the slaves up independent of clagd.
2. also added post_down clagd addon to cleanup the clag-id on ifdown.

9 years ago[Change by Roopa] To prevent ifupdown2 setting clag-bond slave states on ifup.
anuradhak [Mon, 2 Feb 2015 16:46:47 +0000 (08:46 -0800)]
[Change by Roopa] To prevent ifupdown2 setting clag-bond slave states on ifup.

Ticket: CM-4802
Reviewed By: CCR-2507
Testing Done: Yes

1. For clag bonds the slaves are kept down under various conditions (when clagd
is holding the bond in an errDisabled state). ifupdown2 should not set the slave
states for clag bonds.
2. Bonding driver sets the slave states on enslave and release (also for
non-clag bonds) so ifupdown2 doesn't need to set the slave state.

9 years agos/portautoedge/portadminedge/ + configure BA on uplink port
Roopa Prabhu [Thu, 29 Jan 2015 01:06:51 +0000 (17:06 -0800)]
s/portautoedge/portadminedge/ + configure BA on uplink port

Ticket:
Reviewed By:
Testing Done:

9 years agoHandle blank lines correctly in addons.conf
Roopa Prabhu [Thu, 22 Jan 2015 20:56:56 +0000 (12:56 -0800)]
Handle blank lines correctly in addons.conf

Ticket: (Reported by gospo)
Reviewed By:
Testing Done: Tested with blank lines in the file

9 years agoFix typo in comment in default config file
Roopa Prabhu [Fri, 16 Jan 2015 17:54:19 +0000 (09:54 -0800)]
Fix typo in comment in default config file

Ticket:
Reviewed By: trivial
Testing Done: no code change

9 years agoAdded clag-id to the interfaces example file.
anuradhak [Mon, 12 Jan 2015 21:44:54 +0000 (13:44 -0800)]
Added clag-id to the interfaces example file.

Ticket:
Reviewed By:
Testing Done:

9 years agochange ifupdown example for vxlan-learning to the non-default option
Balakrishnan Raman [Wed, 7 Jan 2015 06:12:08 +0000 (22:12 -0800)]
change ifupdown example for vxlan-learning to the non-default option

Ticket: CM-4723
Reviewed By: CCR-2408
Testing Done:

Change example to non-default option - "vxlan-learning off"

9 years agoFix BCM SDK tunnel termination object cleanup on vxlan tunnel delete
Balakrishnan Raman [Wed, 7 Jan 2015 05:16:39 +0000 (21:16 -0800)]
Fix BCM SDK tunnel termination object cleanup on vxlan tunnel delete

Ticket: CM-4723
Reviewed By: CCR-2408
Testing Done: vxlan_fld_test.py:test_bond_vxlinux_convergence_flaps, manual testing

On tunnel delete, tunnel termination object in BCM sdk is not cleaned properly.
As a result, mpls_entry in the hardware is left around even though software has
deleted the tunnel. This prevents subsequent dynamic creation of the same tunnel
and also mac learning, as mpls_entry points to invalid SVP with no CML setting
to do hardware learning.

Also, changes in ifupdown2 to make "vxlan-learning on" as default, so that
user/script doesn't have to explicitly turn on the learning.

9 years agoMove disable ipv6 error to info in mstpctl module (test scripts are
Roopa Prabhu [Sun, 4 Jan 2015 06:35:45 +0000 (22:35 -0800)]
Move disable ipv6 error to info in mstpctl module (test scripts are
still using mstpctl-ports)

Ticket: CM-4622
Reviewed By:
Testing Done: Tested bridge sanity

9 years agoDocument ifreload new option + add it into init.d networking script
Roopa Prabhu [Sat, 3 Jan 2015 08:27:35 +0000 (00:27 -0800)]
Document ifreload new option + add it into init.d networking script

Ticket:
Reviewed By: trivial
Testing Done: Tested build/install/boot + ifupdown2 sanity

9 years agoWarn when bond-min-links is absent or set to '0'
Roopa Prabhu [Sat, 3 Jan 2015 02:20:15 +0000 (18:20 -0800)]
Warn when bond-min-links is absent or set to '0'

Ticket: CM-4327
Reviewed By:
Testing Done: Tested bond config without the bond-min-links attribute

9 years agoAlways set lacp-bypass-allow value to default if lacp-bypass-allow is
Roopa Prabhu [Fri, 2 Jan 2015 04:43:10 +0000 (20:43 -0800)]
Always set lacp-bypass-allow value to default if lacp-bypass-allow is
not given by the user

Ticket: CM-4552
Reviewed By:
Testing Done: Tested with and without lacp-bypass-allow in the file

9 years agoMoved warning to info level
Roopa Prabhu [Thu, 1 Jan 2015 07:47:18 +0000 (23:47 -0800)]
Moved warning to info level

Ticket: CM-4622
Reviewed By:
Testing Done: Tested ifup/ifdown on a bridge

If the file is not present, we will just log it.

9 years agoAdd hack to flap macvlan dev if it already exists and if the route to the vrr network...
Roopa Prabhu [Thu, 1 Jan 2015 07:43:27 +0000 (23:43 -0800)]
Add hack to flap macvlan dev if it already exists and if the route to the vrr network is pointing to it (part of commit 1ef86ab1930d198b0fba7ac33576c22f34d89f14)

Ticket: CM-4639
Reviewed By:
Testing Done: Tested with interfaces file and steps from the bug

This is a Hack to make sure the primary address
is the first in the routing table. Its not a full proof solution.

We use `ip route get` on the vrr network to see which
device the kernel returns. if it is the mac vlan device,
flap the macvlan device to adjust the routing table entry.
flapping the macvlan device makes sure the macvlan
connected route goes through delete + add, hence adjusting
the order in the routing table.

9 years agoAdd hack to flap macvlan dev if it already exists and if the route to
Roopa Prabhu [Thu, 1 Jan 2015 06:51:46 +0000 (22:51 -0800)]
Add hack to flap macvlan dev if it already exists and if the route to
the vrr network is pointing to it

Ticket: CM-4639
Reviewed By:
Testing Done: Tested with interfaces file and steps from the bug

This is a Hack to make sure the primary address
is the first in the routing table. Its not a full proof solution.

We use `ip route get` on the vrr network to see which
device the kernel returns. if it is the mac vlan device,
flap the macvlan device to adjust the routing table entry.
flapping the macvlan device makes sure the macvlan
connected route goes through delete + add, hence adjusting
the order in the routing table.

9 years agoFix link_type flag for objects read from state file + fix link_type
Roopa Prabhu [Wed, 31 Dec 2014 21:58:25 +0000 (13:58 -0800)]
Fix link_type flag for objects read from state file + fix link_type
initializations in a few other places

Ticket: CM-4610
Reviewed By:
Testing Done: Tested with the failing sequence of events described in
the bug

9 years agoMove link_master_slave flag overrides to /sbin/ifupdown + log a msg to
Roopa Prabhu [Tue, 30 Dec 2014 06:07:49 +0000 (22:07 -0800)]
Move link_master_slave flag overrides to /sbin/ifupdown + log a msg to
indicate link_master_slave flag status

Ticket:
Reviewed By:
Testing Done: ifupdown2 sanity tests with link_master_slave feature on/off

9 years agoMoved bond-clag-id attr setting from ifenslave.py to addons/clagd.py
anuradhak [Mon, 29 Dec 2014 19:48:45 +0000 (11:48 -0800)]
Moved bond-clag-id attr setting from ifenslave.py to addons/clagd.py

Ticket:
Reviewed By:
Testing Done:

<DETAILED DESCRIPTION (REPLACE)>
(cherry picked from commit ed008ea9ec176e01b2043b872ad0b8b193f31d0f)

9 years agoChanges to replace clagid setting in the bonding driver with -
anuradhak [Sun, 28 Dec 2014 15:07:47 +0000 (07:07 -0800)]
Changes to replace clagid setting in the bonding driver with -
1. clag_enable flag in the bonding driver
2. And clagid in clagd

Ticket: CM-4091
Reviewed By:
Testing Done: yes

1. Two new commands have been introduced in clagctl - setclagid and showclagid.
These commands will be used by ifupdown2 to set and read/check the clagid.
2. In addition a bit will be maintained per-bond in the bonding driver to
indicate if a bond is a CLAG bond.
(cherry picked from commit cb160bad45d330fe5f544d0d9a40c1c77b9b563f)

9 years agoUpdated in response to review comments in CCR-2301
anuradhak [Mon, 15 Dec 2014 23:32:26 +0000 (15:32 -0800)]
Updated in response to review comments in CCR-2301

Ticket:
Reviewed By:
Testing Done:

<DETAILED DESCRIPTION (REPLACE)>
(cherry picked from commit 48276ee247e230fdd9cd89a33e3b21f13fe85e58)

9 years agoCLAG id for 2.5.0
anuradhak [Tue, 2 Dec 2014 19:32:53 +0000 (11:32 -0800)]
CLAG id for 2.5.0

Ticket:
Reviewed By:
Testing Done: Yes

1. ifupdown2/bonding-driver changes to show/store clag-id (errDisabled/dormant states for bond slaves will follow in 2.5.1)
2. clagd changes for
   a. --forceDynamic - clagid is ignored in this mode
   b. (default) static mode - clagid is matched along with partner mac for dualConnection; also ad_sys_mac is only changed for clag bonds (bonds with a non-zero clag-id).

Conflicts:
patches/kernel/series

9 years agoAdded a few more try-except clauses during bridge config to continue on
Roopa Prabhu [Sun, 28 Dec 2014 04:29:44 +0000 (20:29 -0800)]
Added a few more try-except clauses during bridge config to continue on
failures

Ticket: CM-4382
Reviewed By:
Testing Done: Tested with bridge port addition failure example in the
bug report

9 years agoSuppress 'Network down' warnings when link_master_slave feature is on
Roopa Prabhu [Fri, 26 Dec 2014 17:39:52 +0000 (09:39 -0800)]
Suppress 'Network down' warnings when link_master_slave feature is on

Ticket: CM-4462
Reviewed By:
Testing Done: Tested with interfaces file given in the bug.

The 'network down' msg from the kernel is when the lower device is not
'admin up'. In CM-4462 it is seen when the vlan interface on the bond
is 'admin up' when the bond is still in 'admin down' state.
The bond is also a bridge port so, bond will be 'admin up' when
the bridge it belongs to is brought up.

As link_master_slave feature is on only when all network interfaces
are brought up/down, the states of all interfaces will eventually
converge to 'admin up', so ignoring such transient 'network down' messages.

9 years agoFix error msg strings from rtnetlink module
Roopa Prabhu [Fri, 26 Dec 2014 17:37:42 +0000 (09:37 -0800)]
Fix error msg strings from rtnetlink module

Ticket:
Reviewed By:
Testing Done:

9 years agoFix a bug with disabling of config option link_master_slave + turn off
Roopa Prabhu [Fri, 26 Dec 2014 17:21:50 +0000 (09:21 -0800)]
Fix a bug with disabling of config option link_master_slave + turn off
link master slave feature when '-a' is not specified.

Ticket: CM-4408
Reviewed By:
Testing Done: Tested with ifupdown2.conf link_master_slave on/off

This will make the behaviour change mostly invisible to users

Conflicts:
packages/ifupdown2/ifupdown/ifupdownmain.py

9 years agoAdd rtnetlink api's for 'bridge vlan' and modify 'bridge vlan' commands
Roopa Prabhu [Fri, 26 Dec 2014 17:09:47 +0000 (09:09 -0800)]
Add rtnetlink api's for 'bridge vlan' and modify 'bridge vlan' commands
in vlan module to use rtnetlink api

Ticket: CM-4173
Reviewed By:
Testing Done: Tested new bridge driver svi up/down

This does not move all 'bridge vlan' commands to rtnetlink yet.

9 years agoRework bridge upperiface handling (to fix some problems with the new
Roopa Prabhu [Wed, 24 Dec 2014 20:14:30 +0000 (12:14 -0800)]
Rework bridge upperiface handling (to fix some problems with the new
bridge driver and svi config)

Ticket: CM-3346
Reviewed By:
Testing Done: Tested upper interface handling during ifup of
bonds/bridges/svis

9 years agoMerge branch 'release-2.5' into CumulusLinux-2.5_br
John Berezovik [Wed, 24 Dec 2014 20:20:53 +0000 (12:20 -0800)]
Merge branch 'release-2.5' into CumulusLinux-2.5_br

Conflicts:
build-config/make/local-pkgs.make
packages/cafe/clobjects.py
packages/cafe/clobjects_l3_quagga.py
packages/cl-basefiles/etc/cumulus/datapath/traffic.conf
packages/cl-utilities/etc/monit/monitrc.d/cumulus
packages/clag/ifupdown/clagd.py
packages/ifupdown2/addons/bridge.py
packages/ifupdown2/addons/ifenslave.py
packages/ifupdown2/ifupdown/iface.py
packages/ifupdown2/ifupdown/ifupdownmain.py
packages/ifupdown2/ifupdown/scheduler.py
packages/portwd/portwd
patches/bcm-sdk/series
patches/iproute/bridge-increase-netlink-rcv-buf-size.patch
patches/iproute/series
patches/kernel/driver-support-sff-8436-eeprom.patch
patches/kernel/network-bridge-dont-install-local-mac.patch
patches/kernel/series
patches/lldpd/series
patches/mstpd/clag_port_fsm_reset.patch
patches/mstpd/mstpd_init_system.patch
patches/mstpd/series
patches/quagga/series
rootconf/default/etc/default/cron
switchd/hal.c
switchd/hal_bcm.c
switchd/hal_bcm_config.c
switchd/hal_bcm_config.h
switchd/hal_config.c
switchd/hal_config.h
switchd/hal_int.h
switchd/sync.c
tests/tests/l3/bgp_alpm_test.py
tests/tests/l3/bgp_path_hunting.py
tests/tests/l3/bgp_smoke_test.py
tests/tests/l3/l3_node.py
tests/tests/l3/mc_snoop_test.py
tests/tests/lib/decorators.py
tests/tests/smoke/testifupdown2.py
utilities/ci/pretest.yml
utilities/distro-cmd
utilities/packages/ssim2/setup.py
utilities/packages/ssim2/ssim2/simlib.py

9 years agoAdd check for vlan aware bridge when setting arp_accept to 1
Roopa Prabhu [Wed, 17 Dec 2014 23:09:52 +0000 (15:09 -0800)]
Add check for vlan aware bridge when setting arp_accept to 1

Ticket: CM-4373
Reviewed By:
Testing Done: Tested with address on vlan aware bridge as described in
the bug

9 years agoMultiple fixes and cleanup
Roopa Prabhu [Wed, 17 Dec 2014 20:39:38 +0000 (12:39 -0800)]
Multiple fixes and cleanup

Ticket: CM-3346
Reviewed By:
Testing Done: Tested ifupdown2 sanity

- moved 'admin up' delays that we introduced recently to be
configurable via two ifupdown2.conf attributes
    # Let link master (bridges, bonds) own the link state of slaves
    link_master_slave=1

    # Delay admin state change till the end
    delay_admin_state_change=0

- reduced some redundant traversal of dependency trees

- fixed a few bugs in query check

9 years agofix kernel lacp_fallback_* sysfs file references to use lacp_bypass_*
Roopa Prabhu [Tue, 16 Dec 2014 01:45:49 +0000 (17:45 -0800)]
fix kernel lacp_fallback_* sysfs file references to use lacp_bypass_*

Ticket:
Reviewed By:
Testing Done: Build and sanity test

9 years agobring up bridge ports only after vlan config is done
Roopa Prabhu [Sat, 13 Dec 2014 07:45:31 +0000 (23:45 -0800)]
bring up bridge ports only after vlan config is done

Ticket:
Reviewed By: wkok
Testing Done: unit testing + testing by wkok/purna
(cherry picked from commit 5af90b2893cd54bb21fedaf27ec6dd22de3f2802)

Conflicts:
packages/ifupdown2/addons/bridge.py

9 years agobring up bridge ports only after vlan config is done
Roopa Prabhu [Sat, 13 Dec 2014 07:45:31 +0000 (23:45 -0800)]
bring up bridge ports only after vlan config is done

Ticket:
Reviewed By: wkok
Testing Done: unit testing + testing by wkok/purna

9 years agos/fallback/bypass/
Roopa Prabhu [Sat, 13 Dec 2014 06:56:29 +0000 (22:56 -0800)]
s/fallback/bypass/

Ticket: CM-4452
Reviewed By:
Testing Done: Tested with old and new format

9 years agos/fallback/bypass/
Roopa Prabhu [Sat, 13 Dec 2014 06:56:29 +0000 (22:56 -0800)]
s/fallback/bypass/

Ticket: CM-4452
Reviewed By:
Testing Done: Tested with old and new format

9 years agoMake bridge module the owner of bridge port links (This is to delay
Roopa Prabhu [Fri, 12 Dec 2014 03:06:21 +0000 (19:06 -0800)]
Make bridge module the owner of bridge port links (This is to delay
bridge port up until bridge is up)

Ticket:
Reviewed By:
Testing Done:

9 years agoFix hex flags representation (this was causing wrong evaluation of & and
Roopa Prabhu [Fri, 12 Dec 2014 00:13:25 +0000 (16:13 -0800)]
Fix hex flags representation (this was causing wrong evaluation of & and
| on flags)

Ticket:
Reviewed By:
Testing Done:

Problem seen with ifquery skipping some interfaces because of wrong
evaluation of iface flags

9 years agoAdd a few more trt/except blocks to warn and proceed
Roopa Prabhu [Thu, 11 Dec 2014 05:30:39 +0000 (21:30 -0800)]
Add a few more trt/except blocks to warn and proceed

Ticket:
Reviewed By:
Testing Done:

<DETAILED DESCRIPTION (REPLACE)>
(cherry picked from commit da446f09f0c723852f20ba1c666abcc80b5fe764)

9 years agoAdd a few more trt/except blocks to warn and proceed
Roopa Prabhu [Thu, 11 Dec 2014 05:30:39 +0000 (21:30 -0800)]
Add a few more trt/except blocks to warn and proceed

Ticket:
Reviewed By:
Testing Done:

<DETAILED DESCRIPTION (REPLACE)>

9 years agoRemove stale debug
Roopa Prabhu [Tue, 9 Dec 2014 22:26:11 +0000 (14:26 -0800)]
Remove stale debug

Ticket:
Reviewed By:
Testing Done:
(cherry picked from commit 3ccb3873a995ee7d47134dbbae88ebb424da6d8e)

9 years agoRemove stale debug
Roopa Prabhu [Tue, 9 Dec 2014 22:26:11 +0000 (14:26 -0800)]
Remove stale debug

Ticket:
Reviewed By:
Testing Done:

9 years agoAdd clagd in the default shipped addons.conf and remove the editing of
Roopa Prabhu [Tue, 9 Dec 2014 21:49:23 +0000 (13:49 -0800)]
Add clagd in the default shipped addons.conf and remove the editing of
addons file

Ticket: CM-4423
Reviewed By:
Testing Done: tested ifupdown2 in the presence and absence of clagd

This is a hack to get around an ifupdown2 limitation.
The real fix should come as part of CM-3782
(cherry picked from commit 6fb63d6cdcb660063cf5d07ff28f49c34a306371)