[k8s-R5.0-Single-Yaml-Prov]: k8s provisioning using single yaml failed

Bug #1752657 reported by Pulkit Tandon
14
This bug affects 2 people
Affects Status Importance Assigned to Milestone
Juniper Openstack
Status tracked in Trunk
R5.0
Fix Committed
Critical
Sachchidanand Vaidya
Trunk
In Progress
Critical
Sachchidanand Vaidya

Bug Description

BUG Tempelate:

Configuration:
K8s 1.9.2
Docker version 17.03.1-ce
Centos -7.4

Setup:
3 node setup.
1 Kube master. 1 Controller.
2 Agent+ K8s slaves

Followed steps suggested by Sachin to provision k8s contrail cluster using single yaml provisioning.
Agent containers keep on restarting and do not get spawned.
State of agent containers was:
kube-system contrail-agent-czx2x 1/2 CrashLoopBackOff 7 8m
kube-system contrail-agent-ffmz5 1/2 CrashLoopBackOff 6 8m

There are multiple problem leading to this state:

1. By default in contrail.yaml, the PHYSICAL_INTERFACE: is always set to:
PHYSICAL_INTERFACE: eno1
At that time the agent will give following errors:
2018-03-01 Thu 16:49:30:061.102 UTC nodec58 [Thread 140193993205952, Pid 1]: Error reading interface <eno1>. Error number 19 : No such device
2018-03-01 Thu 16:54:33:048.405 UTC nodec58 [Thread 140017763444928, Pid 1]: Error reading interface <eno1>. Error number 19 : No such device

2. To correct this, this line can be deleted manually from contrail.yaml and the file can be loaded again.
This resulted in losing the connectivity to agents as they lose default routes.
Workaround to this is adding the default routes manually.

3. After above problems are corrected, it lands up in an issue to connect to the API server.
ERROR Tue Feb 27 06:53:43 UTC 2018: Config API servers 10.204.217.108 are not responding on port 8082. Exiting...
WARNING Tue Feb 27 06:53:43 UTC 2018: Some of Config API servers 10.204.217.108 are not responding on port 8082.
Provisioning: provision_vrouter.py --oper add --host_name nodec60 --host_ip 10.204.217.100: 0/10
Traceback (most recent call last):
  File “/opt/contrail/utils/provision_vrouter.py”, line 274, in <module>
    main()
  File “/opt/contrail/utils/provision_vrouter.py”, line 270, in main
    VrouterProvisioner(args_str)
  File “/opt/contrail/utils/provision_vrouter.py”, line 35, in __init__
    api_server_use_ssl=self._args.api_server_use_ssl)
  File “/opt/contrail/utils/vnc_admin_api.py”, line 35, in __init__
    auth_host=self.auth_host)
  File “/usr/lib/python2.7/site-packages/vnc_api/vnc_api.py”, line 460, in __init__
    retry_on_error=False)
  File “/usr/lib/python2.7/site-packages/vnc_api/vnc_api.py”, line 943, in _request
    raise ConnectionError
requests.exceptions.ConnectionError
Provisioning: provision_vrouter.py --oper add --host_name nodec60 --host_ip 10.204.217.100: 1/10
Traceback (most recent call last):
  File “/opt/contrail/utils/provision_vrouter.py”, line 274, in <module>
    main()
  File “/opt/contrail/utils/provision_vrouter.py”, line 270, in main
    VrouterProvisioner(args_str)
  File “/opt/contrail/utils/provision_vrouter.py”, line 35, in __init__
    api_server_use_ssl=self._args.api_server_use_ssl)
  File “/opt/contrail/utils/vnc_admin_api.py”, line 35, in __init__
    auth_host=self.auth_host)
  File “/usr/lib/python2.7/site-packages/vnc_api/vnc_api.py”, line 460, in __init__
    retry_on_error=False)
  File “/usr/lib/python2.7/site-packages/vnc_api/vnc_api.py”, line 943, in _request
    raise ConnectionError
requests.exceptions.ConnectionError

Revision history for this message
OpenContrail Admin (ci-admin-f) wrote : [Review update] master

Review in progress for https://review.opencontrail.org/40827
Submitter: Sachchidanand Vaidya (<email address hidden>)

Jeba Paulaiyan (jebap)
tags: added: sanityblocker
Revision history for this message
Sachchidanand Vaidya (vaidyasd) wrote :

Moving all single yaml provisioning jobs in this bug:
Single Yaml Provisioning
1. Single controller and multiple agents
2. Multiple controller with distributed micro services and multiple agents. Single Interface
3. Multiple controller with distributed micro services and multiple agents. Multi Interface(Control/Data interface)

Revision history for this message
OpenContrail Admin (ci-admin-f) wrote :

Review in progress for https://review.opencontrail.org/41961
Submitter: Sachchidanand Vaidya (<email address hidden>)

Revision history for this message
OpenContrail Admin (ci-admin-f) wrote : [Review update] R5.0

Review in progress for https://review.opencontrail.org/42157
Submitter: Sachchidanand Vaidya (<email address hidden>)

Revision history for this message
OpenContrail Admin (ci-admin-f) wrote : A change has been merged

Reviewed: https://review.opencontrail.org/42157
Committed: http://github.com/Juniper/contrail-container-builder/commit/2e9347d57b127d3afdf749ab40d17b28935815da
Submitter: Zuul v3 CI (<email address hidden>)
Branch: R5.0

commit 2e9347d57b127d3afdf749ab40d17b28935815da
Author: Sachchidanand Vaidya <email address hidden>
Date: Mon Apr 16 08:13:42 2018 -0700

Changes to make single-yaml provisioning work for Kubernetes
- Add analytics zookeeper as DaemonSet.
- Separate ConfigMap for config & analytics zookeeper port.
- Change taint.
- yaml based of latest template that includes contrail-status support

Change-Id: I446680ec11ccca49ff888351af55b22cc93ad57e
Closes-bug: #1752657

Jeba Paulaiyan (jebap)
tags: added: sanity
removed: sanityblocker
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.