Public bug reported:
In our openstack stein installation (neutron 14.4.2) we upgraded keepalived
from 1.3.9 to 2.2.4.
After that when restarting the neutron-l3-agent we saw that the router state of
routers with external gatways were not able to update anymore. So we ended up
with only standby routers even though keepalived is working fine and we can see
one keepalived in master state.
After debugging a bit we found the following traceback:
```
Traceback (most recent call last):
File "/usr/lib/python3/dist-packages/eventlet/hubs/hub.py", line 461, in
fire_timers
timer()
File "/usr/lib/python3/dist-packages/eventlet/hubs/timer.py", line 59, in
__call__
cb(*args, **kw)
File "/usr/lib/python3/dist-packages/neutron/agent/l3/ha.py", line 166, in
_enqueue_state_change
ri.set_external_gw_port_link_status(link_up=True, set_gw=True)
File "/usr/lib/python3/dist-packages/neutron/agent/l3/ha_router.py", line
547, in set_external_gw_port_link_status
ns_name, preserve_ips)
File "/usr/lib/python3/dist-packages/neutron/agent/l3/router_info.py", line
750, in _external_gateway_settings
clean_connections=True)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/interface.py", line
179, in init_router_port
preserve_ips)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/interface.py", line
203, in set_onlink_routes
device.route.add_onlink_route(route)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
662, in add_onlink_route
self.add_route(cidr, scope='link')
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
712, in add_route
self._run_as_root_detect_device_not_found([ip_version], args)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
615, in _run_as_root_detect_device_not_found
raise exceptions.DeviceNotFoundError(device_name=self.name)
File "/usr/lib/python3/dist-packages/oslo_utils/excutils.py", line 220, in
__exit__
self.force_reraise()
File "/usr/lib/python3/dist-packages/oslo_utils/excutils.py", line 196, in
force_reraise
six.reraise(self.type_, self.value, self.tb)
File "/usr/local/lib/python3.6/dist-packages/six.py", line 719, in reraise
raise value
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
610, in _run_as_root_detect_device_not_found
return self._as_root(options, tuple(args))
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
407, in _as_root
use_root_namespace=use_root_namespace)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
121, in _as_root
namespace=namespace)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
129, in _execute
log_fail_as_error=self.log_fail_as_error)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/utils.py", line 147,
in execute
returncode=returncode)
neutron_lib.exceptions.ProcessExecutionError: Exit code: 2; Stdin: ; Stdout: ;
Stderr: RTNETLINK answers: File exists
```
This traceback is triggered because the routers got duplicated routes, here an
example router:
```
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip r
default via x.x.244.67 dev qg-6c2ee5e0-ad proto 18
default via x.x.244.67 dev qg-6c2ee5e0-ad
10.0.0.0/24 dev qr-15d63a29-8e proto kernel scope link src 10.0.0.10
169.254.0.0/24 dev ha-f64d319f-ed proto kernel scope link src 169.254.0.13
169.254.192.0/18 dev ha-f64d319f-ed proto kernel scope link src 169.254.197.12
x.x.244.64/26 dev qg-6c2ee5e0-ad proto kernel scope link src x.x.244.116
x.x.244.128/25 dev qg-6c2ee5e0-ad proto 18 scope link
x.x.244.128/25 dev qg-6c2ee5e0-ad scope link
```
First I thought that we hit a keepalived bug which I filed here:
https://github.com/acassen/keepalived/issues/2076
What I got to understand from the communication with pqarmitage from the issue
is that keepalived is setting `proto 18/keepalived` in newer versions and I
think that this breaks with neutron.
So what I assume is happening here is the following.
On a "fresh" router or a failover the qg- interface of the router is down,
therefore keepalived is not able to set the virtual routes. neutron then
creates the gateway routes through the set_external_gw_port_link_status
function
(https://opendev.org/openstack/neutron/src/tag/14.4.2/neutron/agent/l3/ha_router.py#L528)
after it brings up the qg- interface. When I now restart the neutron-l3-agent
it reloads keepalived which triggers keepalived to recreate the virtual_routes
which it was not able to create when the qg- interface was down and because of
the new functionality it creates the same route but with the addition of `proto
18` and we end up with duplicated routes. After that neutron fails on the `ip
route replace` command with the RTNETLINK answers: File exists error.
Here is a router example of getting into this state:
backup router:
```
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group
default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
1755: ha-f64d319f-ed: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue
state UNKNOWN group default qlen 1000
link/ether fa:16:3e:56:3f:59 brd ff:ff:ff:ff:ff:ff
inet 169.254.197.12/18 brd 169.254.255.255 scope global ha-f64d319f-ed
valid_lft forever preferred_lft forever
inet6 fe80::f816:3eff:fe56:3f59/64 scope link
valid_lft forever preferred_lft forever
1756: qr-15d63a29-8e: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue
state UNKNOWN group default qlen 1000
link/ether fa:16:3e:a9:cc:49 brd ff:ff:ff:ff:ff:ff
1757: qg-6c2ee5e0-ad: <BROADCAST,MULTICAST> mtu 1500 qdisc noqueue state DOWN
group default qlen 1000
link/ether fa:16:3e:d2:6c:c7 brd ff:ff:ff:ff:ff:ff
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip r
169.254.192.0/18 dev ha-f64d319f-ed proto kernel scope link src 169.254.197.12
```
After a failover to the backup node where I assume neutron is setting the
gateway routes:
```
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group
default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
1755: ha-f64d319f-ed: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue
state UNKNOWN group default qlen 1000
link/ether fa:16:3e:56:3f:59 brd ff:ff:ff:ff:ff:ff
inet 169.254.197.12/18 brd 169.254.255.255 scope global ha-f64d319f-ed
valid_lft forever preferred_lft forever
inet 169.254.0.13/24 scope global ha-f64d319f-ed
valid_lft forever preferred_lft forever
inet6 fe80::f816:3eff:fe56:3f59/64 scope link
valid_lft forever preferred_lft forever
1756: qr-15d63a29-8e: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue
state UNKNOWN group default qlen 1000
link/ether fa:16:3e:a9:cc:49 brd ff:ff:ff:ff:ff:ff
inet 10.0.0.10/24 scope global qr-15d63a29-8e
valid_lft forever preferred_lft forever
inet6 fe80::f816:3eff:fea9:cc49/64 scope link nodad
valid_lft forever preferred_lft forever
1757: qg-6c2ee5e0-ad: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue
state UNKNOWN group default qlen 1000
link/ether fa:16:3e:d2:6c:c7 brd ff:ff:ff:ff:ff:ff
inet x.x.244.116/26 scope global qg-6c2ee5e0-ad
valid_lft forever preferred_lft forever
inet6 x.x:1003::22b/64 scope global nodad
valid_lft forever preferred_lft forever
inet6 fe80::f816:3eff:fed2:6cc7/64 scope link nodad
valid_lft forever preferred_lft forever
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip r
default via x.x.244.67 dev qg-6c2ee5e0-ad
10.0.0.0/24 dev qr-15d63a29-8e proto kernel scope link src 10.0.0.10
169.254.0.0/24 dev ha-f64d319f-ed proto kernel scope link src 169.254.0.13
169.254.192.0/18 dev ha-f64d319f-ed proto kernel scope link src 169.254.197.12
x.x.244.64/26 dev qg-6c2ee5e0-ad proto kernel scope link src x.x.244.116
x.x.244.128/25 dev qg-6c2ee5e0-ad scope link
```
And then after a neutron-l3-agent restart which triggers a keepalived reload:
```
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip r
default via x.x.244.67 dev qg-6c2ee5e0-ad proto 18
default via x.x.244.67 dev qg-6c2ee5e0-ad
10.0.0.0/24 dev qr-15d63a29-8e proto kernel scope link src 10.0.0.10
169.254.0.0/24 dev ha-f64d319f-ed proto kernel scope link src 169.254.0.13
169.254.192.0/18 dev ha-f64d319f-ed proto kernel scope link src 169.254.197.12
x.x.244.64/26 dev qg-6c2ee5e0-ad proto kernel scope link src x.x.244.116
x.x.244.128/25 dev qg-6c2ee5e0-ad proto 18 scope link
x.x.244.128/25 dev qg-6c2ee5e0-ad scope link
```
** Affects: neutron
Importance: Undecided
Status: New
--
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1956846
Title:
ha router duplicated routes
Status in neutron:
New
Bug description:
In our openstack stein installation (neutron 14.4.2) we upgraded keepalived
from 1.3.9 to 2.2.4.
After that when restarting the neutron-l3-agent we saw that the router state
of routers with external gatways were not able to update anymore. So we ended
up with only standby routers even though keepalived is working fine and we can
see one keepalived in master state.
After debugging a bit we found the following traceback:
```
Traceback (most recent call last):
File "/usr/lib/python3/dist-packages/eventlet/hubs/hub.py", line 461, in
fire_timers
timer()
File "/usr/lib/python3/dist-packages/eventlet/hubs/timer.py", line 59, in
__call__
cb(*args, **kw)
File "/usr/lib/python3/dist-packages/neutron/agent/l3/ha.py", line 166, in
_enqueue_state_change
ri.set_external_gw_port_link_status(link_up=True, set_gw=True)
File "/usr/lib/python3/dist-packages/neutron/agent/l3/ha_router.py", line
547, in set_external_gw_port_link_status
ns_name, preserve_ips)
File "/usr/lib/python3/dist-packages/neutron/agent/l3/router_info.py", line
750, in _external_gateway_settings
clean_connections=True)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/interface.py",
line 179, in init_router_port
preserve_ips)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/interface.py",
line 203, in set_onlink_routes
device.route.add_onlink_route(route)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
662, in add_onlink_route
self.add_route(cidr, scope='link')
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
712, in add_route
self._run_as_root_detect_device_not_found([ip_version], args)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
615, in _run_as_root_detect_device_not_found
raise exceptions.DeviceNotFoundError(device_name=self.name)
File "/usr/lib/python3/dist-packages/oslo_utils/excutils.py", line 220, in
__exit__
self.force_reraise()
File "/usr/lib/python3/dist-packages/oslo_utils/excutils.py", line 196, in
force_reraise
six.reraise(self.type_, self.value, self.tb)
File "/usr/local/lib/python3.6/dist-packages/six.py", line 719, in reraise
raise value
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
610, in _run_as_root_detect_device_not_found
return self._as_root(options, tuple(args))
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
407, in _as_root
use_root_namespace=use_root_namespace)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
121, in _as_root
namespace=namespace)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/ip_lib.py", line
129, in _execute
log_fail_as_error=self.log_fail_as_error)
File "/usr/lib/python3/dist-packages/neutron/agent/linux/utils.py", line
147, in execute
returncode=returncode)
neutron_lib.exceptions.ProcessExecutionError: Exit code: 2; Stdin: ; Stdout:
; Stderr: RTNETLINK answers: File exists
```
This traceback is triggered because the routers got duplicated routes, here
an example router:
```
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip r
default via x.x.244.67 dev qg-6c2ee5e0-ad proto 18
default via x.x.244.67 dev qg-6c2ee5e0-ad
10.0.0.0/24 dev qr-15d63a29-8e proto kernel scope link src 10.0.0.10
169.254.0.0/24 dev ha-f64d319f-ed proto kernel scope link src 169.254.0.13
169.254.192.0/18 dev ha-f64d319f-ed proto kernel scope link src 169.254.197.12
x.x.244.64/26 dev qg-6c2ee5e0-ad proto kernel scope link src x.x.244.116
x.x.244.128/25 dev qg-6c2ee5e0-ad proto 18 scope link
x.x.244.128/25 dev qg-6c2ee5e0-ad scope link
```
First I thought that we hit a keepalived bug which I filed here:
https://github.com/acassen/keepalived/issues/2076
What I got to understand from the communication with pqarmitage from the
issue is that keepalived is setting `proto 18/keepalived` in newer versions and
I think that this breaks with neutron.
So what I assume is happening here is the following.
On a "fresh" router or a failover the qg- interface of the router is down,
therefore keepalived is not able to set the virtual routes. neutron then
creates the gateway routes through the set_external_gw_port_link_status
function
(https://opendev.org/openstack/neutron/src/tag/14.4.2/neutron/agent/l3/ha_router.py#L528)
after it brings up the qg- interface. When I now restart the neutron-l3-agent
it reloads keepalived which triggers keepalived to recreate the virtual_routes
which it was not able to create when the qg- interface was down and because of
the new functionality it creates the same route but with the addition of `proto
18` and we end up with duplicated routes. After that neutron fails on the `ip
route replace` command with the RTNETLINK answers: File exists error.
Here is a router example of getting into this state:
backup router:
```
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group
default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
1755: ha-f64d319f-ed: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc
noqueue state UNKNOWN group default qlen 1000
link/ether fa:16:3e:56:3f:59 brd ff:ff:ff:ff:ff:ff
inet 169.254.197.12/18 brd 169.254.255.255 scope global ha-f64d319f-ed
valid_lft forever preferred_lft forever
inet6 fe80::f816:3eff:fe56:3f59/64 scope link
valid_lft forever preferred_lft forever
1756: qr-15d63a29-8e: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc
noqueue state UNKNOWN group default qlen 1000
link/ether fa:16:3e:a9:cc:49 brd ff:ff:ff:ff:ff:ff
1757: qg-6c2ee5e0-ad: <BROADCAST,MULTICAST> mtu 1500 qdisc noqueue state DOWN
group default qlen 1000
link/ether fa:16:3e:d2:6c:c7 brd ff:ff:ff:ff:ff:ff
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip r
169.254.192.0/18 dev ha-f64d319f-ed proto kernel scope link src 169.254.197.12
```
After a failover to the backup node where I assume neutron is setting the
gateway routes:
```
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group
default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
1755: ha-f64d319f-ed: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc
noqueue state UNKNOWN group default qlen 1000
link/ether fa:16:3e:56:3f:59 brd ff:ff:ff:ff:ff:ff
inet 169.254.197.12/18 brd 169.254.255.255 scope global ha-f64d319f-ed
valid_lft forever preferred_lft forever
inet 169.254.0.13/24 scope global ha-f64d319f-ed
valid_lft forever preferred_lft forever
inet6 fe80::f816:3eff:fe56:3f59/64 scope link
valid_lft forever preferred_lft forever
1756: qr-15d63a29-8e: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc
noqueue state UNKNOWN group default qlen 1000
link/ether fa:16:3e:a9:cc:49 brd ff:ff:ff:ff:ff:ff
inet 10.0.0.10/24 scope global qr-15d63a29-8e
valid_lft forever preferred_lft forever
inet6 fe80::f816:3eff:fea9:cc49/64 scope link nodad
valid_lft forever preferred_lft forever
1757: qg-6c2ee5e0-ad: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc
noqueue state UNKNOWN group default qlen 1000
link/ether fa:16:3e:d2:6c:c7 brd ff:ff:ff:ff:ff:ff
inet x.x.244.116/26 scope global qg-6c2ee5e0-ad
valid_lft forever preferred_lft forever
inet6 x.x:1003::22b/64 scope global nodad
valid_lft forever preferred_lft forever
inet6 fe80::f816:3eff:fed2:6cc7/64 scope link nodad
valid_lft forever preferred_lft forever
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip r
default via x.x.244.67 dev qg-6c2ee5e0-ad
10.0.0.0/24 dev qr-15d63a29-8e proto kernel scope link src 10.0.0.10
169.254.0.0/24 dev ha-f64d319f-ed proto kernel scope link src 169.254.0.13
169.254.192.0/18 dev ha-f64d319f-ed proto kernel scope link src 169.254.197.12
x.x.244.64/26 dev qg-6c2ee5e0-ad proto kernel scope link src x.x.244.116
x.x.244.128/25 dev qg-6c2ee5e0-ad scope link
```
And then after a neutron-l3-agent restart which triggers a keepalived reload:
```
ip netns exec qrouter-76f69b0d-c9ac-4a98-851a-f74b23b2de49 ip r
default via x.x.244.67 dev qg-6c2ee5e0-ad proto 18
default via x.x.244.67 dev qg-6c2ee5e0-ad
10.0.0.0/24 dev qr-15d63a29-8e proto kernel scope link src 10.0.0.10
169.254.0.0/24 dev ha-f64d319f-ed proto kernel scope link src 169.254.0.13
169.254.192.0/18 dev ha-f64d319f-ed proto kernel scope link src 169.254.197.12
x.x.244.64/26 dev qg-6c2ee5e0-ad proto kernel scope link src x.x.244.116
x.x.244.128/25 dev qg-6c2ee5e0-ad proto 18 scope link
x.x.244.128/25 dev qg-6c2ee5e0-ad scope link
```
To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1956846/+subscriptions
--
Mailing list: https://launchpad.net/~yahoo-eng-team
Post to : [email protected]
Unsubscribe : https://launchpad.net/~yahoo-eng-team
More help : https://help.launchpad.net/ListHelp
