Hi Joseph,

Below is my patch to make rabbitmq pod start in ipv6 setup.(This is not a formal patch)

From 99fc5aff8d70645c3130552789ea51d82cd287b2 Mon Sep 17 00:00:00 2001

From: Zhipeng Liu <zhipengs.liu@intel.com>

Date: Thu, 19 Mar 2020 18:53:07 +0800

Subject: [PATCH] Fix rabbtimq pod not starting issue in IPv6 setup

Signed-off-by: Zhipeng Liu <zhipengs.liu@intel.com>

---

rabbitmq/templates/configmap-etc.yaml    | 8 +++++++-

rabbitmq/templates/job-cluster-wait.yaml | 2 ++

rabbitmq/templates/statefulset.yaml      | 8 ++++++++

3 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/rabbitmq/templates/configmap-etc.yaml b/rabbitmq/templates/configmap-etc.yaml

index 87f25f5..18e2145 100644

--- a/rabbitmq/templates/configmap-etc.yaml

+++ b/rabbitmq/templates/configmap-etc.yaml

@@ -21,7 +21,8 @@ limitations under the License.

{{- $_ := print "kubernetes.default.svc." $envAll.Values.endpoints.cluster_domain_suffix | set $envAll.Values.conf.rabbitmq.cluster_formation.k8s "host" -}}

{{- end -}}

-{{- $_ := print "0.0.0.0:" ( tuple "oslo_messaging" "internal" "amqp" . | include "helm-toolkit.endpoints.endpoint_port_lookup") | set $envAll.Values.conf.rabbitmq.listeners.tcp "1" -}}

+{{- $_ := print ":::" ( tuple "oslo_messaging" "internal" "amqp" . | include "helm-toolkit.endpoints.endpoint_port_lookup") | set $envAll.Values.conf.rabbitmq.listeners.tcp "1" -}}

+

 ---

apiVersion: v1

@@ -33,4 +34,9 @@ data:

{{ tuple "etc/_enabled_plugins.tpl" . | include  "helm-toolkit.utils.template" | indent 4 }}

   rabbitmq.conf: |

{{ include "rabbitmq.utils.to_rabbit_config" $envAll.Values.conf.rabbitmq | indent 4 }}

+  rabbitmq-env.conf: |

+    SERVER_ADDITIONAL_ERL_ARGS="+A 128 -kernel inetrc '/etc/rabbitmq/erl_inetrc' -proto_dist inet6_tcp"

+    CTL_ERL_ARGS="-proto_dist inet6_tcp"

+  erl_inetrc: |

+    {inet6, true}.

{{ end }}

diff --git a/rabbitmq/templates/job-cluster-wait.yaml b/rabbitmq/templates/job-cluster-wait.yaml

index bf8e710..24a30a3 100644

--- a/rabbitmq/templates/job-cluster-wait.yaml

+++ b/rabbitmq/templates/job-cluster-wait.yaml

@@ -66,6 +66,8 @@ spec:

           env:

             - name: RABBITMQ_ADMIN_CONNECTION

               value: {{ tuple "oslo_messaging" "internal" "user" "http" $envAll | include "helm-toolkit.endpoints.authenticated_endpoint_uri_lookup" | quote }}

+            - name: RABBITMQ_CTL_ERL_ARGS

+              value: "-proto_dist inet6_tcp"

             - name: RABBIT_REPLICA_COUNT

               value: {{ $envAll.Values.pod.replicas.server | quote }}

           command:

diff --git a/rabbitmq/templates/statefulset.yaml b/rabbitmq/templates/statefulset.yaml

index 41dc198..64fc1cf 100644

--- a/rabbitmq/templates/statefulset.yaml

+++ b/rabbitmq/templates/statefulset.yaml

@@ -228,6 +228,14 @@ spec:

               mountPath: /etc/rabbitmq/rabbitmq.conf

               subPath: rabbitmq.conf

               readOnly: true

+            - name: rabbitmq-etc

+              mountPath: /etc/rabbitmq/rabbitmq-env.conf

+              subPath: rabbitmq-env.conf

+              readOnly: true

+            - name: rabbitmq-etc

+              mountPath: /etc/rabbitmq/erl_inetrc

+              subPath: erl_inetrc

+              readOnly: true

       volumes:

         - name: pod-tmp

           emptyDir: {}

--

2.7.4

From: Richard, Joseph <Joseph.Richard@windriver.com>
Sent: 2020年3月19日 0:35
To: Liu, ZhipengS <zhipengs.liu@intel.com>; starlingx-discuss@lists.starlingx.io
Subject: Re: [Starlingx-discuss] IPv6 issue when apply OpenStack on IPv6 simplex.

Have you tried changing envAll.Values.conf.rabbitmq.listeners.tcp?

I'd need to see the changes you've made to get rabbitmq running to provide any other ideas.

---

From: Liu, ZhipengS <zhipengs.liu@intel.com>
Sent: Wednesday, March 18, 2020 10:53 AM
To: Richard, Joseph <Joseph.Richard@windriver.com>; starlingx-discuss@lists.starlingx.io <starlingx-discuss@lists.starlingx.io>
Subject: RE: [Starlingx-discuss] IPv6 issue when apply OpenStack on IPv6 simplex.

Hi Joseph,

I will submit one patch for helm chart change later after fixing current issue.

BTW, Seems we have no rabbitmqadmin in rabbitmq pod but rabbitmq-cluster-wait job.

I have done below test and have some conclusion now.

1) Rabbtimq already started normally, and start listen on [::]:5672/15672/25672 port

2) It seems 15672 port access blocked for ipv6 (5672 is OK). In ipv4 simplex setup, same test below are all OK. Any idea?

rabbitmq@osh-openstack-rabbitmq-rabbitmq-0:/usr/local$ rabbitmqctl status

Status of node rabbit@osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local ...

Runtime

OS PID: 303

OS: Linux

Uptime (seconds): 6752

RabbitMQ version: 3.8.3

Node name: rabbit@osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local

Erlang configuration: Erlang/OTP 22 [erts-10.7] [source] [64-bit] [smp:6:6] [ds:6:6:10] [async-threads:128]

Erlang processes: 426 used, 1048576 limit

Scheduler run queue: 1

Cluster heartbeat timeout (net_ticktime): 60

Plugins

Enabled plugin file: /etc/rabbitmq/enabled_plugins

Enabled plugins:

* rabbitmq_management

* rabbitmq_web_dispatch

* rabbitmq_peer_discovery_k8s

* rabbitmq_peer_discovery_common

* rabbitmq_management_agent

* amqp_client

* cowboy

* cowlib

Totals

Connection count: 0

Queue count: 0

Virtual host count: 0

Listeners

Interface: [::], port: 25672, protocol: clustering, purpose: inter-node and CLI tool communication

Interface: [::], port: 5672, protocol: amqp, purpose: AMQP 0-9-1 and AMQP 1.0

Interface: [::], port: 15672, protocol: http, purpose: HTTP API

controller-0:~$ wget http://rabbit@osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local:15672/

--2020-03-18 14:37:45--  http://rabbit@osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local:15672/

Resolving osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local (osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local)... fd03::1e6d:6abf

Connecting to osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local (osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local)|fd03::1e6d:6abf|:15672... failed: Connection refused. //same as we see in the log of

Rabbitmq-cluster-wait pod.

controller-0:~$ wget http://rabbit@osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local:5672/

--2020-03-18 14:37:53--  http://rabbit@osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local:5672/

Resolving osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local (osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local)... fd03::1e6d:6abf

Connecting to osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local (osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local)|fd03::1e6d:6abf|:5672... connected.

controller-0:~$ wget http://[fd03::1e6d:6abf]:15672/

--2020-03-18 14:38:56--  http://[fd03::1e6d:6abf]:15672/

Connecting to [fd03::1e6d:6abf]:15672... failed: Connection refused.

controller-0:~$ wget http://[fd03::1e6d:6abf]:5672/

--2020-03-18 14:39:00--  http://[fd03::1e6d:6abf]:5672/

Connecting to [fd03::1e6d:6abf]:5672... connected.

Thanks!

Zhipeng

From: Richard, Joseph <Joseph.Richard@windriver.com>
Sent: 2020年3月18日 21:48
To: Liu, ZhipengS <zhipengs.liu@intel.com>; starlingx-discuss@lists.starlingx.io
Subject: Re: [Starlingx-discuss] IPv6 issue when apply OpenStack on IPv6 simplex.

Can you open a review with the change you made to fix the nxdomain issue?

Is rabbitmqadmin able to connect if you run it from within osh-openstack-rabbitmq-rabbitmq-0?

I see from your logs that the osh-openstack-rabbitmq-cluster-wait is from before rabbitmq was running.  Can you see anything in the rabbitmq when the cluster-wait pod fails?

---

From: Liu, ZhipengS <zhipengs.liu@intel.com>
Sent: Tuesday, March 17, 2020 9:46 PM
To: Richard, Joseph <Joseph.Richard@windriver.com>; starlingx-discuss@lists.starlingx.io <starlingx-discuss@lists.starlingx.io>
Subject: RE: [Starlingx-discuss] IPv6 issue when apply OpenStack on IPv6 simplex.

Hi Joseph,

After did some changes in rabbitmq helm chart, I have fixed nxdomain issue in my ipv6 simplex setup.

Below two pods can run up.

osh-openstack-rabbitmq-cluster-wait-9rw6p 1/1 Running 0 17m
osh-openstack-rabbitmq-rabbitmq-0 1/1 Running 0 17m

However, still have another issue.
In pod osh-openstack-rabbitmq-cluster-wait, it will use rabbitmqadmin to connect rabbitmq,
but always get error. 

++ active_rabbit_nodes
2020-03-17T10:31:12.124589385Z stderr F ++ wc -w
2020-03-17T10:31:12.134367271Z stderr F ++ rabbitmqadmin_authed list nodes -f bash
2020-03-17T10:31:12.134427089Z stderr F ++ set +x
2020-03-17T10:31:12.179073378Z stderr F Traceback (most recent call last):
2020-03-17T10:31:12.179644557Z stderr F error: [Errno 111] Connection refused
2020-03-17T10:31:12.17964969Z stderr F *** Could not connect: [Errno 111] Connection refused

In the log of rabbitmq node, I can see below.

2020-03-17T15:40:27.762196613Z stdout F 2020-03-17 15:40:27.761 [info] <0.495.0> started TCP listener on [::]:5672

2020-03-17T15:40:27.807886117Z stdout F 2020-03-17 15:40:27.807 [info] <0.552.0> Management plugin: HTTP (non-TLS) listener started on port 15672

2020-03-17T15:40:27.895551288Z stdout F 2020-03-17 15:40:27.895 [info] <0.8.0> Server startup complete; 5 plugins started.

2020-03-17T15:40:27.895555144Z stdout F  * rabbitmq_peer_discovery_k8s

2020-03-17T15:40:27.895557288Z stdout F  * rabbitmq_management

2020-03-17T15:40:27.895559593Z stdout F  * rabbitmq_management_agent

2020-03-17T15:40:27.895561807Z stdout F  * rabbitmq_web_dispatch

2020-03-17T15:40:27.895564619Z stdout F  * rabbitmq_peer_discovery_common

I’m still working on this issue. Any comment or proposal, thanks!

Thanks!

Zhipeng

From: Liu, ZhipengS
Sent: 2020年3月16日 16:57
To: 'Richard, Joseph' <Joseph.Richard@windriver.com>; 'starlingx-discuss@lists.starlingx.io' <starlingx-discuss@lists.starlingx.io>
Subject: RE: [Starlingx-discuss] IPv6 issue when apply OpenStack on IPv6 simplex.

Hi Joseph,

I tried many methods, but still not fixed the issue.

I found there are some related issued fixed in upstream, but not yet merged to formal release.

I’m discussing with them. Do you have any comment?

https://github.com/rabbitmq/rabbitmq-peer-discovery-k8s/issues/55

Thanks!

Zhipeng

From: Liu, ZhipengS
Sent: 2020年3月12日 9:53
To: 'Richard, Joseph' <Joseph.Richard@windriver.com>; starlingx-discuss@lists.starlingx.io
Subject: RE: [Starlingx-discuss] IPv6 issue when apply OpenStack on IPv6 simplex.

Hi Joseph,

Thank you so much! 😊

I will work further based on your provided clue.

Best Regards,

Zhipeng

From: Richard, Joseph <Joseph.Richard@windriver.com>
Sent: 2020年3月12日 5:06
To: Liu, ZhipengS <zhipengs.liu@intel.com>; starlingx-discuss@lists.starlingx.io
Subject: Re: [Starlingx-discuss] IPv6 issue when apply OpenStack on IPv6 simplex.

I see a similar issue online here:

https://groups.google.com/forum/#!topic/rabbitmq-users/9ir1WK-nYBA

epmd fails to start in IPv6 environment Posted 7/15/19 9:43 PM, 12 messages groups.google.com

I expect that you are going to have to update rabbmitmq docker images to 3.7.16(currently 3.7.13), and then modify the helm plugin to pass in these values when using ipv6:

SERVER_ADDITIONAL_ERL_ARGS='-proto_dist inet6_tcp'

CTL_ERL_ARGS='-proto_dist inet6_tcp'

Not sure if this is possible right now, or if you will need to modify openstack-helm-infra/rabbitmq in order to be able to pass these environment variables in.

---

From: Liu, ZhipengS <zhipengs.liu@intel.com>
Sent: Thursday, February 27, 2020 8:00 AM
To: starlingx-discuss@lists.starlingx.io <starlingx-discuss@lists.starlingx.io>
Subject: [Starlingx-discuss] IPv6 issue when apply OpenStack on IPv6 simplex.

Hi all,

For below LP IPv6: stx-openstack app apply failed by time out waiting for the condition "grpc_status"

https://bugs.launchpad.net/starlingx/+bug/1859641

I provided a patch which can fix the known issue in LP.

However, it still failed later when starting rabbitmq pod.

controller-0:~$ kubectl -n openstack get po
NAME READY STATUS RESTARTS AGE
ingress-c74489c96-jsb7n 1/1 Running 0 17h
ingress-error-pages-5889bfcf6c-h8kl6 1/1 Running 0 17h
mariadb-ingress-b5cd65f4b-sq46c 1/1 Running 0 17h
mariadb-ingress-error-pages-6d8b79b74f-9zmpc 1/1 Running 0 17h
mariadb-server-0 1/1 Running 0 17h
osh-openstack-memcached-memcached-545668bdbd-9jkv2 1/1 Running 0 17h
osh-openstack-rabbitmq-cluster-wait-wbnwp 0/1 Init:0/2 0 17h
osh-openstack-rabbitmq-rabbitmq-0 0/1 CrashLoopBackOff 207 17h

There are below error logs

2020-02-26T07:43:36.451103975Z stderr F ++ echo osh-openstack-rabbitmq-rabbitmq-0
2020-02-26T07:43:36.45142814Z stderr F ++ awk -F - '{print $NF}'
2020-02-26T07:43:36.454160452Z stderr F + POD_INCREMENT=0
2020-02-26T07:43:36.454757022Z stderr F + '[' 0 -eq 0 ']'
2020-02-26T07:43:36.454795721Z stderr F + exec rabbitmq-server
2020-02-26T07:43:36.849021551Z stdout F ERROR: epmd error for host osh-openstack-rabbitmq-rabbitmq-0.rabbitmq.openstack.svc.cluster.local: nxdomain (non-existing domain)

In /etc/hosts file below, I just found localhost4, no localhost6.  Is it expected?

I added below line manually after the line of 127.0.0.1. However, it didn’t work☹

::1             localhost       localhost.localdomain localhost6 localhost6.localdomain6

# HEADER: This file was autogenerated at 2020-02-26 09:00:35 +0000
# HEADER: by puppet. While it can still be managed manually, it
# HEADER: is definitely not recommended.
127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4
fd01::1 controller
169.254.202.1 pxecontroller
fd00::3 oamcontroller
fd01::4 controller-platform-nfs
fd01::3 controller-1
fd01::2 controller-0
fd02::3 controller-1-cluster-host
fd02::2 controller-0-cluster-host
fd01::e8b6:66fe storage-0

I saw that in ./stx-puppet/puppet-manifests/src/modules/platform/manifests/config.pp:215

class platform::config::hosts

  inherits ::platform::config::params {

  # The localhost should resolve to the IPv4 loopback address only, therefore

# ensure the IPv6 address is removed from configured hosts

  resources { 'host': purge => true }

  $localhost = {

    'localhost' => {

      ip => '127.0.0.1',

      host_aliases => ['localhost.localdomain', 'localhost4', 'localhost4.localdomain4']

    },

  }

  $merged_hosts = merge($localhost, $hosts)

  create_resources('host', $merged_hosts, {})

}

Any one can give some help?

Thanks!
Zhipeng