Couldn't get docker plugin to work

@hongbin, I'm trying to re-produce this problem.

while use latest devstack-plugin-container install docker, I get the follwoing error:
root@devstack:/vagrant# uname -a
Linux devstack 4.4.0-77-generic #98-Ubuntu SMP Wed Apr 26 08:34:02 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux
root@devstack:/vagrant# curl -fsSL https://download.docker.com/linux/ubuntu/gpg
curl: (6) Could not resolve host: download.docker.com

Looks like I get similar problem with:
http://www.ttlsa.com/linux/curl-6-couldnt-resolve-host/

When I add '-4' it works for me(use ipv4):
root@devstack:/vagrant# curl -fsSL4 https://download.docker.com/linux/ubuntu/gpg

Did you get same problem before?
my vm is using : ubuntu/xenial64 20170503.1.0

Thanks, I will skip this problem manually to go on research your problem this week.

Hi Liping,

I didn't encounter this problem. Here is my environment:

$ uname -a
Linux testkuryr 4.4.0-79-generic #100-Ubuntu SMP Wed May 17 19:58:14 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

However, welcome to patch the devstack-plugin-container if you have a chance .

@hongbin, I can reproduce this problem, I find the problem is when run "iface.add_ip('fd34:d27c:33d3:0:f816:3eff:fe3f:8693', 64)"

If I only run iface.add_ip IPV4, it works:
>>> import pyroute2
>>>
>>> ip = pyroute2.IPDB()
>>> with ip.interfaces['tap1e510214-51'] as iface:
... iface.add_ip('10.0.0.8', 26)
...
{'family': 0, 'txqlen': 1000, 'ipdb_scope': 'system', 'index': 20, 'operstate': 'LOWERLAYERDOWN', 'num_tx_queues': 1, 'group': 0, 'carrier_changes': 3, 'ipaddr': (('fe80::fcda:a6ff:fe93:e328', 64), ('10.0.0.8', 26)), 'neighbours': (), 'ifname': 'tap1e510214-51', 'promiscuity': 0, 'linkmode': 0, 'broadcast': 'ff:ff:ff:ff:ff:ff', 'link': 19, 'address': 'fe:da:a6:93:e3:28', 'vlans': (), 'ipdb_priority': 0, 'kind': 'veth', 'qdisc': 'noqueue', 'mtu': 1500, 'num_rx_queues': 1, 'carrier': 0, 'flags': 3, 'ifi_type': 1, 'proto_down': 0, 'ports': ()}
>>>

If I add IPV6, it failed

>>> import pyroute2
>>>
>>> ip = pyroute2.IPDB()
>>> with ip.interfaces['tap1e510214-51'] as iface:
... #iface.add_ip('10.0.0.8', 26)
... iface.add_ip('fd34:d27c:33d3:0:f816:3eff:fe3f:8693', 64)
...
{'family': 0, 'txqlen': 1000, 'ipdb_scope': 'system', 'index': 20, 'operstate': 'LOWERLAYERDOWN', 'num_tx_queues': 1, 'group': 0, 'carrier_changes': 3, 'ipaddr': (('fe80::fcda:a6ff:fe93:e328', 64), ('10.0.0.8', 26)), 'neighbours': (), 'ifname': 'tap1e510214-51', 'promiscuity': 0, 'linkmode': 0, 'broadcast': 'ff:ff:ff:ff:ff:ff', 'link': 19, 'address': 'fe:da:a6:93:e3:28', 'vlans': (), 'ipdb_priority': 0, 'kind': 'veth', 'qdisc': 'noqueue', 'mtu': 1500, 'num_rx_queues': 1, 'carrier': 0, 'flags': 3, 'ifi_type': 1, 'proto_down': 0, 'ports': ()}
Traceback (most recent call last):
  File "<stdin>", line 3, in <module>
  File "/usr/lib/python2.7/site-packages/pyroute2/ipdb/transactional.py", line 209, in __exit__
    self.commit()
  File "/usr/lib/python2.7/site-packages/pyroute2/ipdb/interfaces.py", line 1057, in commit
    raise error
pyroute2.ipdb.exceptions.CommitException: ipaddr target is not set

@hongbin,
BTW, same command can success when I run directly on host, it failed when I run in plugin container. Need to further debug.

@hongbin,

Here is what I find right now.

I use the following code run at host directly and in plugin container, it success in host, but failed in plugin container.

Here is code:
root@devstack:~# cat a.py
import pyroute2

ip = pyroute2.IPDB()
with ip.interfaces['tap1e510214-51'] as iface:
    iface.add_ip('fd34:d27c:33d3:0:f816:3eff:fe3f:8693/64')

I find if I run the code in side plugin container, I see the IPV6 is setted on interface, but it is fd34:d27c:33d3::f816:3eff:fe3f:8693%32687/64, it is very strange that it has "%32687" there.

Here is output of ifconfig:
tap1e510214-51 Link encap:Ethernet HWaddr FE:DA:A6:93:E3:28
          inet6 addr: fd34:d27c:33d3::f816:3eff:fe3f:8693%32687/64 Scope:Global
          UP BROADCAST MTU:1500 Metric:1
          RX packets:8 errors:0 dropped:0 overruns:0 frame:0
          TX packets:7 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:816 (816.0 B) TX bytes:738 (738.0 B)

I need to go on check why it has "%32687" in the IPV6 address.

Hi Liping,

Thanks for tracing down the issue. I can confirm that all the tap interfaces has the strange characters sequence:

# ifconfig tapce248b27-59
tapce248b27-59 Link encap:Ethernet HWaddr 1E:22:16:02:F0:B5
          inet6 addr: fe80::1c22:16ff:fe02:f0b5%32733/64 Scope:Link
          UP BROADCAST MTU:1500 Metric:1
          RX packets:9 errors:0 dropped:0 overruns:0 frame:0
          TX packets:7 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:926 (926.0 B) TX bytes:738 (738.0 B)

@hongbin,

Sorry for the delay, after debug, I find this is a bug[1] in pyroute 0.4.18, it has been fixed in 0.4.19. I manually upgrade pyroute in pluginv2 container, it works for me with your script:

root@devstack:/home/vagrant/devstack# docker run -itd --net=testnet --ip=$IP --ip6=$IP6 nginx
a3f6d81e7776633347bf81711f43ff59d7f9725ed13926356475f68e094e6d63

[1]https://github.com/svinota/pyroute2/commit/e576eb45a750bd9f0049dfe25a0cc714fefbb988

Regards,
Liping Mao

@Hongbin,

Since the latest pyroute2 is 0.4.19 now（Released on 2017-08-01), the pyroute2 in global requirements is > 0.4.17, so can you try rebuild the pluginv2 image? 0.4.19 should already fix the problem.

Not sure if we need to update in global requirement, something like:
pyroute2 > 0.4.17 != 0.4.18

Thoughts?

Regards,
Liping Mao

Hi Liping,

I can confirm that plugin-rebuild can resolve the original error. The newly built plugin has 0.4.19 so it looks the problem is possibly related to 0.4.18. Thanks for your trouble-shooting and providing this information. I think it is a good idea to ban 0.4.18 in global requirements, but the requirement repo is freeze now.

After resolving the original issue, it seems I can go further but hit another issue on port binding:

  $ bash create_from_private_net.sh
  WARNING: setting legacy OS_TENANT_NAME to support cli tools.
  ae94fc83bc729d6a3731e2305c5b8dd78173b03f0f2530a68594c1ce8de2935b
  neutron CLI is deprecated and will be removed in the future. Use openstack CLI instead.
  c7aea0c39c501317a6a951a4f73a546c1ce3028b67766332cb15307ca27f987a
  docker: Error response from daemon: failed to create endpoint zealous_fermat on network testnet:
  NetworkDriver.CreateEndpoint: vif_type(ovs) is not supported. A binding script for this type can't be found.

I can look into the issue once I have a chance. Please let me know if you happen to know how to resolve that error as well :). Appreciate your help.

Hi Hongbin,

This is a configuration problem, since pluginv2 will load configuration in /etc/kuryr/kuryr.conf .
Please double check , it should be:
[DEFAULT]
bindir = /usr/libexec/kuryr

The default value in ubuntu is bindir = /usr/local/libexec/kuryr, this do not works in pluginv2, since in the image, the bind scripts are in /usr/libexec/kuryr.

You need to update kuryr.conf, and restart pluginv2 . This should fix your problem.

Regards,
Liping Mao

Thanks Liping,

After fixing the config file, it works very well.