Docker Community Forums

Share and learn in the Docker community.

Resolved: Docker Cloud assigns 10.7.0.0/16 to containers but Weave uses 10.128.0.0/16


(Erezny) #1

On a BYON I have three containers, InfluxDB, Grafana, and HAProxy. They cannot communicate with each other using the overlay network. I cannot figure out how to either configure weave for the correct ip range or to configure dockercloud to assign the correct ip range to containers. Is there a dockercloud agent config parameter I can set to fix this?

HAProxy sees this in its ENV,
“GRAFANA_DASBOB_PROD_ENV_DOCKERCLOUD_IP_ADDRESS=10.7.0.5/16”,

Weave logs
Warning: ‘–connlimit’ is deprecated, it will be replaced by ‘–conn-limit’ soon. See usage.
INFO: 2016/02/16 16:20:21.948150 Command line options: map[init-peer-count:2 name:52:59:c3:c7:f4:77 no-dns:true port:6783 datapath:datapath dns-effective-listen-address:172.17.0.1 http-addr:127.0.0.1:6784 ipalloc-range:10.128.0.0/10 nickname:Zapp no-discovery:true trusted-subnets: conn-limit:0]
INFO: 2016/02/16 16:20:21.948238 Command line peers: []
INFO: 2016/02/16 16:20:22.056677 Communication between peers via untrusted networks is encrypted.
INFO: 2016/02/16 16:20:22.057445 Our name is 52:59:c3:c7:f4:77(Zapp)
INFO: 2016/02/16 16:20:22.058858 Docker API on unix///var/run/docker.sock: &[KernelVersion=3.19.0-49-generic BuildTime=Mon Nov 30 21:56:07 UTC 2015 Version=1.9.1-cs2 ApiVersion=1.21 GitCommit=4ade326 GoVersion=go1.4.3 Os=linux Arch=amd64]
INFO: 2016/02/16 16:20:22.058984 Sniffing traffic on datapath (via ODP)
INFO: 2016/02/16 16:20:22.060088 Listening for HTTP control messages on 127.0.0.1:6784
INFO: 2016/02/16 16:20:22.253758 [allocator 52:59:c3:c7:f4:77] Ignored address 10.7.255.254 claimed by weave:expose - not in our universe
INFO: 2016/02/16 16:20:24.847982 [allocator 52:59:c3:c7:f4:77] Ignored address 10.7.255.254 claimed by weave:expose - not in our universe
INFO: 2016/02/16 16:20:47.859701 ->[52.90.141.19:54660] connection accepted
INFO: 2016/02/16 16:20:47.912084 ->[52.90.141.19:54660|42:98:81:dc:57:5d(ip-10-42-0-33)]: connection ready; using protocol version 2
INFO: 2016/02/16 16:20:47.912134 Unable to use fastdp for connection to 42:98:81:dc:57:5d(ip-10-42-0-33): encryption not supported
INFO: 2016/02/16 16:20:47.912216 overlay_switch ->[42:98:81:dc:57:5d(ip-10-42-0-33)] using sleeve
INFO: 2016/02/16 16:20:47.912356 ->[52.90.141.19:54660|42:98:81:dc:57:5d(ip-10-42-0-33)]: connection added (new peer)
INFO: 2016/02/16 16:20:48.053844 ->[52.90.141.19:54660|42:98:81:dc:57:5d(ip-10-42-0-33)]: connection fully established
INFO: 2016/02/16 16:20:48.054543 EMSGSIZE on send, expecting PMTU update (IP packet was 60052 bytes, payload was 60044 bytes)
INFO: 2016/02/16 16:20:48.378560 sleeve ->[52.90.141.19:6783|42:98:81:dc:57:5d(ip-10-42-0-33)]: Effective MTU verified at 1414
INFO: 2016/02/16 16:21:02.834102 [allocator 52:59:c3:c7:f4:77] Ignored address 10.7.0.5 claimed by 64107c49ba1d220c6522729d231b61029699205c1393490472dda3e022ac6903 - not in our universe

References:
github.com/tutumcloud/weave-daemon/issues/34


(Fernando Mayo) #3

The IP range Docker Cloud uses for the overlay network is 10.7.0.0/16. We don’t use weave’s IPAM module so 10.128.0.0/16 is not used at all.

I checked your node and it seems that some containers kept restarting.

In order to further troubleshoot this, can you please “exec” into the containers, print the list of interfaces and and try to ping each other?


(Borja Burgos) #4