https://bugzilla.redhat.com/show_bug.cgi?id=1065001
Bug ID: 1065001
Summary: Starting docker service removes IPv6 default route
Product: Fedora EPEL
Version: el6
Component: docker-io
Assignee: lsm5(a)redhat.com
Reporter: redhat(a)subs.maneos.org
QA Contact: extras-qa(a)fedoraproject.org
CC: admiller(a)redhat.com, golang(a)lists.fedoraproject.org,
lsm5(a)redhat.com, mattdm(a)redhat.com,
mgoldman(a)redhat.com, skottler(a)redhat.com,
vbatts(a)redhat.com
Description of problem:
The docker init script does this:
preexec="/sbin/sysctl"
[ -x $preexec ] || exit 6
$preexec -w net.ipv4.ip_forward=1 > /dev/null 2>&1
$preexec -w net.ipv6.conf.all.forwarding=1 > /dev/null 2>&1
These two settings are not quite equivalent. In particular,
net.ipv6.conf.all.forwarding=1 disables autoconfiguration.
Version-Release number of selected component (if applicable):
0.7.6-2.el6
How reproducible:
Steps to Reproduce:
1. Start with a host using stateless IPV6 networking configuration.
2. Install docker-io.
3. Start docker-io service.
Actual results:
Loss of ::/0 route (previously auto-configured via a router advertisment).
Expected results:
No changes in routing table.
Additional info:
According to Documentation/networking/ip-sysctl.txt setting
net.ipv6.conf.xxx.accept_ra to 2 should re-enable it for that interface but
that doesn't seem to work for me.
--
You are receiving this mail because:
You are on the CC list for the bug.