Bug 1136337

+++ This bug was initially created as a clone of Bug #1136016 +++

Description of problem:


Version-Release number of selected component (if applicable):


How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results:


Additional info:

--- Additional comment from john.haxby on 2014-09-01 15:25:32 CEST ---

[I'm not sure how that initial empty description got created.  Sorry.]

Description of problem:

If you create a bond whose name is a substring of an existing bond, then the new bond will not be created.  This is because install_bonding_driver() in /etc/sysconfig/network-scripts/network-functions contains the test "grep -sq "$1" /sys/class/net/bonding_masters".

Version-Release number of selected component (if applicable):
    initscripts-9.51-2.fc20.x86_64


How reproducible: Always.  Easily reproduced with a slightly contrived example.


Steps to Reproduce:
1.  Create /etc/sysconfig/network-/scripts/ifcfg-bond1 and ifcfg-bond11 like this:

    DEVICE=bond1
    TYPE=Bond
    BONDING_MASTER=yes
    BOOTPROTO=none
    ONBOOT=yes
    BONDING_OPTS="miimon=100 mode=802.3ad"

ifcfg-bond11 is the same except that the DEVCIE is, of course, bond11.

2. ifup bond11
3. ifup bond1

Actual results:

bond11 comes up, bond1 reports errors:

./network-functions: line 439: /sys/class/net/bond1/bonding/slaves: No such file or directory
./network-functions: line 439: [: too many arguments
./network-functions: line 439: /sys/class/net/bond1/bonding/slaves: No such file or directory
./network-functions: line 439: [: too many arguments
ERROR    : [/etc/sysconfig/network-scripts/ifup-eth] Device bond1 does not seem to be present, delaying initialisation.


Expected results: both bonds come up


Additional info:

As I say, this is a contrived example.   Interfaces are normally brought up in lexical order so bond1 would come up first and bond11 afterwards.   The fix is easy and not even worth a patch file:   change "grep -sq" to "fgrep -sqx"

--- Additional comment from john.haxby on 2014-09-01 16:26:29 CEST ---

I noticed that commit e956710ffda24a569460688f99a1d410328bbc61 is a partial fix for this problem.  If uses "grep -sqw" but there's no reason why I shouldn't call my bonds bond1 and bond1.89fe (that's closer to the original problem I had).

Changing to fgrep -x both avoids that problem and the possible problem with regexp matches.

--- Additional comment from john.haxby on 2014-09-01 16:47:34 CEST ---

This is a patch against the git master rather than the current F20 version.   It doesn't need to change much to become a patch against F20.