Jump to: navigation, search

Difference between revisions of "StarlingX/Containers/Limitations"

(Added workaround for router/dhcp scheduling.)
Line 4: Line 4:
 
           - Tracked by https://bugs.launchpad.net/starlingx/+bug/1815541 and https://bugs.launchpad.net/starlingx/+bug/1814595
 
           - Tracked by https://bugs.launchpad.net/starlingx/+bug/1815541 and https://bugs.launchpad.net/starlingx/+bug/1814595
 
2. Autonomous reboots of controller are not yet fully supported.  Containers may hang while the controller is rebooting.  They likely will recover after the controller recovers.  Manual recovery may be required.   
 
2. Autonomous reboots of controller are not yet fully supported.  Containers may hang while the controller is rebooting.  They likely will recover after the controller recovers.  Manual recovery may be required.   
 +
          - For AIO-DX, if the controller with mariadb pod is rebooted, containers will be down until the controller recovers. 
 +
          - If the controller dies and does not recover then the containers will not recover. 
 +
          - This is currently working to address this via SB https://storyboard.openstack.org/#!/story/2004712
  
 
3. IPv6 networking is not yet supported.
 
3. IPv6 networking is not yet supported.

Revision as of 18:42, 6 March 2019

Limitations for a containers configuration as of Feb 28/2019:

1. Containers may hang after several hours or days running, in particular rabbit container, due to kernel jbd thread hanging.

          - Tracked by https://bugs.launchpad.net/starlingx/+bug/1815541 and https://bugs.launchpad.net/starlingx/+bug/1814595

2. Autonomous reboots of controller are not yet fully supported. Containers may hang while the controller is rebooting. They likely will recover after the controller recovers. Manual recovery may be required.

          - For AIO-DX, if the controller with mariadb pod is rebooted, containers will be down until the controller recovers.  
          - If the controller dies and does not recover then the containers will not recover.  
          - This is currently working to address this via SB https://storyboard.openstack.org/#!/story/2004712

3. IPv6 networking is not yet supported.

4. HTTPS configurations are not yet supported.

5. Distributed Cloud and Regions configurations are not yet supported.

6. Neutron routers/dhcp servers do not move on a lock/unlock for AIO-DX or lock/unlock of compute host -- likely means networking down until host recovers or routers/dhcp servers moved manually.

neutron l3-agent-router-remove [-h] L3_AGENT ROUTER
neutron l3-agent-router-add [-h] L3_AGENT ROUTER
neutron dhcp-agent-network-add [-h] DHCP_AGENT NETWORK
neutron dhcp-agent-network-remove [-h] DHCP_AGENT NETWORK