Page tree
Skip to end of metadata
Go to start of metadata

 Meetings for High Availability Project

Meeting on 13:00-14:00 UTC, Wednesday, Mar. 21

Meeting summary

    1. Georg Kunz (georgk,    13:01:26)
    2. Kanglin (Kanglin,    13:01:35)
    3. Dimitrios Markou (mardim,    13:01:37)
  1. HA test case update (fuqiao, 13:01:58)
    1. ODL noHA testcase is merged (mardim,    13:02:44)
    2. we would like to go through all the test cases    we have so far. See which cases are ready for merge. and ask Ross to    merge certain project before next week (fuqiao,    13:02:54)
    3. ODL HA testcase I have a draft (mardim,    13:03:14)
    4.     (mardim,    13:03:26)
    5. TC087 ODL noHA test case is merged (fuqiao,    13:03:46)
    6. TO merge this in we should fix apex integration    with yardstick (mardim,    13:03:59)
    7. we need apex pod.yaml for that (mardim,    13:04:17)
    8. ACTION: mardim will    fix the Apex integration with yardstick (mardim,    13:04:37)
    9. TC085 & TC086 can't merge for F, because of    lack of installer supporting VM HA and Hypervisor HA. (fuqiao,    13:05:33)
    10. VM HA feature comes out with Queen release. and    we are currently working with compass and XCI to make it integrated    for G release. therefor TC085 & TC086 probabally should    targeting for G (fuqiao,    13:07:19)
    11. TC090 & TC091 ready for merge (fuqiao,    13:07:37)
    12. ACTION: Kanglin will    reach out to Ross to have TC090 & TC091 merge before next    week. (fuqiao,    13:12:19)
    13.     (georgk,    13:15:12)
    14. for Fraser release, TC085 TC090 TC091 are ready    to merge. other test cases have issue with passing CI (fuqiao,    13:15:37)
    15. ACTION: fuqiao will    submit release exception for the other test cases (fuqiao,    13:16:11)
    16. ACTION: Fuqiao will    look into the HA analysis doc and merge it before this week     (fuqiao,    13:33:34)
    17.     (georgk,    13:40:26)

Meeting ended at 13:46:57 UTC (full logs).

Action items

  1. mardim will fix the Apex integration with yardstick
  2. Kanglin will reach out to Ross to have TC090 & TC091 merge before next week.
  3. fuqiao will submit release exception for the other test cases
  4. Fuqiao will look into the HA analysis doc and merge it before this week


Meeting on 13:00-14:00 UTC, Wednesday, Feb. 28

IRC Channel: #opnfv-meeting


HA test case progress update


Meeting summary

* fuqiao  (fuqiao, 13:01:28)
* Dimitrios Markou  (mardim, 13:01:40)
* Kanglin  (Kanglin, 13:02:52)
* HA test case progress update  (fuqiao, 13:05:40)
  * the MS6 is March 16. all test cases have to be merged before that if
    we like to be included in the F release  (fuqiao, 13:06:13)
  * Georg Kunz  (georgk, 13:07:00)
  * LINK:  (mardim, 13:11:25)
  * LINK:  (mardim,
  * Dimitrios is now working on SDN HA test cases, but has problem with
    the topology. A jira ticket is raised for that bug  (fuqiao,
  * Wangmeng  (wangmeng, 13:24:21)
  * Kanglin will upload new TC description and TC in a few days
    (fuqiao, 13:26:43)
  * SDN TC link  (fuqiao,
  * Wangmeng will finish the VM HA and hypervisor HA TC before Mar. 16
    (fuqiao, 13:31:58)

Meeting ended at 13:33:45 UTC.

Meeting on 13:30-14:00 UTC, Wednesday, Jan. 24


#opnfv-meeting: high availavility project weekly meeting

Meeting started by fuqiao at 13:29:39 UTC (full logs). 

Meeting summary

    1. fuqiao (fuqiao, 13:29:56)
    2. Georg Kunz (georgk, 13:30:22)
    3. Kanglin (Kanglin, 13:30:36)
    4. Greg Waines (GregWaines, 13:31:59)
  1. HA test case process update (fuqiao, 13:32:06)
    1. (fuqiao, 13:33:06)
    2. (fuqiao, 13:33:28)
    3. (fuqiao, 13:33:49)
    4. Kanglin indicates that the current HA framework can not support packet generation. only support connectivity check with ping (fuqiao, 13:42:52)
    5. ACTIONKanglin will go to Kubi and Mingjiang to make sure whether we can use packet generation tool support by yardstick (fuqiao, 13:43:24)
    6. is a simple software based one that you could run in another VM ... (GregWaines, 13:43:38)
    7. ... actually this might be the better link ... it's DPDK-based packet generator (GregWaines, 13:44:40)
    8. (GregWaines, 13:45:32)
    9. (GregWaines, 13:47:11)
    10. (Kanglin, 13:54:03)
    11. wiki page for yardstick test cases: (fuqiao, 13:54:34)
    12. ACTIONKanglin will add TC into the wiki page for HA-31, HA-33 and HA-34 (fuqiao, 13:56:49)
    13. Kanglin mentioned the L2/L3 VPN connectivity can be verified in VM (fuqiao, 13:58:39)

Meeting ended at 14:01:06 UTC (full logs). 

Action items

  1. Kanglin will go to Kubi and Mingjiang to make sure whether we can use packet generation tool support by yardstick
  2. Kanglin will add TC into the wiki page for HA-31, HA-33 and HA-34

Action items, by person

  1. Kanglin
    1. Kanglin will go to Kubi and Mingjiang to make sure whether we can use packet generation tool support by yardstick
    2. Kanglin will add TC into the wiki page for HA-31, HA-33 and HA-34

People present (lines said)

  1. fuqiao (47)
  2. georgk (17)
  3. Kanglin (15)
  4. GregWaines (11)
  5. collabot (5)
  6. stefan______ (2)

Generated by MeetBot 0.1.4.



Meeting on 13:00-14:00 UTC, Wednesday, Jan. 10


Toll free NO.

USA: +1 (571) 317-3116


Meeting ID: 391-235-029



1, previous test cases overview and update plan for the F release --  Kang Lin, Qiu Juan

  • TC for VIM components: HA for scheduler, conductor, agents, database instances...
  • TC for Controller: Controller shutdown, adding more monitors based on the original case

2, Updates for the SDN related test case – Srikanth

3, Update for the VM HA related test cases – fuqiao


Meeting on 13:00-14:00 UTC, Wednesday, Nov. 1


Toll free NO.

USA: +1 (571) 317-3116


Meeting ID: 391-235-029


F release plan

1, HA API release plan - Greg

2, HA test case release plan. please add comments and suggestions in the etherpad page before the meeting. we will go through the list and make out the release plan for F release.

Meeting on 13:00-14:00 UTC, Wednesday, Oct. 18


1, Georg introduced the on-going work in the QA team of OpenStack. They are planning to work out some HA test cases. Georg suggested it will be good that they adopt the HA test cases developed by OPNFV.

2, Srikanth mentioned there are also some on-going work within Erricsson that they would also like to contribute to OPNFV. Basically includes the following 3 HA test cases 

VM migration
networking robustness of SDN controller
DC Gateway HA

3, Kang Lin and Qiu Juan introduced their plan for the F release. They original plan is to update the Gui and framework of HA testing in Yardstick. Detailed test cases are not decided yet. But they are quite open to suggestions and ideas for future test cases.

4, Yamei introduced the HA test cases in China Mobile, which they also have a team focus on developing automatic test cases and can contribute to OPNFV.

5, The team decide to create an etherpad page to collect the planned test cases for F release, and propose these work to the QA team afterwards.

Meeting on 13:00-14:00 UTC, Wednesday, Jul. 26


1, Greg updates the progress of BP for masakari. The BPs have been accepted a few weeks ago, and code development work has began. We are now targeting Q release of openstack. We plan to integrate the API work into installers of OPNFV after this is integrated in OpenStack releases. Therefore the work may wait till next year.

2, Kanglin introduces the HA testcases for E release. 3 testcases are included, which are TC for pacemaker, rabbitMQ, and virtual router. 

3, Kanglin introduces his acedemic work, which is a map of HA analysis of NFV. This map will help us locate the failure of NFV and the recovery mechenism, which testcase is missing for HA, and also help us better explain the HA requirement. 


Meeting on 13:00-14:00 UTC, Wednesday, Mar. 3rd



The HA team had good discussion with the barometer team. For the Heartbeat API, the Barometer team mentioned they have similar approach, but focus more on collecting metrics in the infrustructure rather than in VM. There could be cooperation between the two teams on this. For the server group API, the Barometer team said they had nothing similar, nor does it conflict with anything barometer does. The team get positive feedback from the barometer team, making sure the proposed schemes do no confict to the barometer schemes. Fuqiao will add the Barometer team as reviewers for the doc commit, and will get the doc reviewed and merged within 2 weeks. The HA team will then move to detailed deployment, and will consider using compass installer to install the proposed schemes. We are targeting E release for the installer.

Meeting on 13:00-14:00 UTC, Wednesday, Apr. 12nd


  • VM Heartbeating & Health-checking
    • libvirt watchdog and its integration / use with OpenStack
    • Update the architecture diagram to be consistent with the most current DOCTOR Architecture Diagrams
      • e.g. Vitrage & Congress are not necessarily deployed at same time
      • use DOCTOR terminology where applicable, e.g. “inspector modules”
      • OPNFV DOCTOR includes patches/components of AODH, NOVA, NEUTRON
        … not just Vitrage and Congress, as shown in diagram
        • (although also had comment that should remove OPNFV DOCTOR outline from diagram as
          OPNFV DOCTOR is a requirement specification and not an implementation)
      • indicate (possibly just in text below diagram) that the “Guest Heartbeat / Health-Check Server” on Controller
        Node is possibly not required, as the Vitrage data source interface can be remotely reached by
        “Guest Heartbeat / Health-Check Compute” on the Compute Node
    • In text, provide a little more detail on content of actual messaging,
      e.g. PDUs and rough message content
    • OVERALL:
      • believe there was general agreement in the way that VM Heartbeating & Health-checking
        was integrated / inter-worked with OPNFV DOCTOR’s Vitrage/Congress and overall OPNFV
        fault reporting architecture
      • key feedback was to understand and highlight the additional value of the VM Heartbeating &
        Health-checking functionality over the existing libvirt watchdog integration into OpenStack
  • Server Group Messaging
    • suggestion that Rabbit MQ pub/sub messaging could be an alternative for routing of messages
      • this is an implementation detail though
    • general discussions on “HA use cases” for how this messaging could be leveraged
      • e.g. split-brain avoidance, faster peer VM state change notifications,
    • Overall:
      • agreement that the Server Group Messaging Architecture did NOT conflict with Doctor Architecture
      • need further review with OPNFV MANO Team as to how they would position this functionality
        • position it as an alternative for various HA use cases ?
        • versus
        • mandating that this service group messaging be used for specific HA user cases ?


Meeting on 13:00-14:00 UTC, Wednesday, Mar. 29th


HA API doc updates – Greg

The team goes through the update version. Some members of the MANO WG join the discussion and provide useful suggestions. Detailed suggestions are as follows:
1) The team agrees to restructure the doc as a proposal, not a total solution.
2) Michael, from the MANO WG, indicates that currently there are more than one framework to improve the availability and reliability of the OPNFV Platform, which could be a good thing. However coordination work is needed in order to tell people how to behave in a unified way.
3) priority should be defined among multiple schemes to avoid race condition
4) The team agrees to include the heart beat part and peer state notification session for the D release.
5) It is suggested to continue the discussion of the proposal with the parameter team and doctor team.

Meeting on 13:00-14:00 UTC, Wednesday, Mar. 8th


The team goes through the updated Overview document. Suggestions are listed below:

1) propose the updated doc to the MANO WG for review and discussion, coordinate the related work.

2) Seperate the heartbeat section into a single document for review, try to finish this section for the D release. Leave the rest of the doc for future release.

3) Add deliveribles for each section.

Meeting on 13:00-14:00 UTC, Wednesday, Feb. 15th


1 HA API doc updates – Greg

The team has discussion about the updated HA API overview documents.

Most of the questions are around who are the consumers for these API. How should these API fixed into the current ETSI NFV flow? Greg will add a few paragragh to include detailed explaination of the relationship between NFVO, VNFM and these APIs. The basic idea is these API should and will be consumed by the NFVO and VNFM through certain nova API. In order to mininize the changes in nova, a nova proxy is used, the NOVA PROXY takes over the NOVA REST API port and intercepts all commands in order to leverage this HOST-to-GUEST API to inform the VM of the event that is about to occur to it, give the VM a chance to reject the command, and then (depending on outcome) forwards the command on to the real NOVA. The team also suggests in this case, VMs not leveraging these APIs should also be considered, so that these APIs can just be an option. Commonds to the VMs not using these APIs should go directly from the nova proxy to the real nova.

Greg will upload a new version of the overview doc and will triger another round of review afterwards.


Meeting on 13:00-14:00 UTC, Wednesday, Feb. 8th


1, HA Test Case Updates and Plan in D Release – Kanglin

Kanglin introduced the progress and release plan for HA testing in yardstick. The progress mainly includes the following 3 aspects:

1) framework improvement

2) Dashboard improvement

3) Testing improvement, adding multuple processes to run so as to get a more accurate outage time.

They also provide a summary for the existing test cases, and a suggestted testing list for future releases. Test cases on virtual VMs are suggested.

fuqiao suggested to add another testcase on pacemaker.

Since MS6 is next week, the team agreed there may not be time for new testcases for D release. The team will begin discussion on future test cases soon after so that can make sure to catch up with the E release.


2. HA API Doc updates – Greg

Greg made updates for the HA API doc. An overview doc is provided, and the original doc is splitted into two seperate doc, one introduce the basic massage in the host, and the other introduce the host - guest API. We didn't have enough time to go through the whole doc. We will have another meeting same time next week to continue the discussion. And also suggested everyone to review the doc first before the next meeting.

Meeting on 13:00-14:00 UTC, Wednesday, Jan. 18th


Wanghui introduced the scope and plan for the Rescuer team. The rescuer team will focus on the date pretection issue, currently mainly in VIM. it will use karbor as its upstream project, and is intended to collect requirement and gap analysis from the OPNFV community. The team has close connection with the OpenStack Karbor project, so can be predicted to work closely with the upsteams. Hui also introduced more details about the Karbor project in OpenStack.

The plan for the D release is to work out requirement for data pretection. The HA team suggested Hui to look into the release doc of the HA project first, and see if they can get more clue for the requirement specificly on data pretection. The HA requirement doc and scenario analysis doc can help to identify requirement and scenarios for data pretection. And  the deployment framework doc can help to understand the HA deployment  context when talking about data pretection.

fuqiao will send out the release doc to hui after the meeting.

Meeting on 13:00-14:00 UTC, Wednesday, Jan. 11th


1 Introduction of Rescuer

Hui didn't show up during the call. Fuqiao has contacted him and another call will be arranged on the next week.

2 HA API doc

Stefan: suggest an overview doc which discribes the complete packet and the overall HA Guest API functionally should be provided before talking about implementation.
Greg: will add that doc
Yuan Yue: asks which parts are provided by KVM, Qume, openstack. What code we should add
Plan to provide mainly the following 3 aspects
1)definition of messaging API between guest and host,
2)Host side change in openstack. Currently not part of any opensource project. Since OPNFV do not manage patches over openstack, we may need to work with Openstack project (e.g. nova) and have these codes accepted.
3) guest side deployment sample, optional peice, can implement based on the spec.
Yuan Yue: modification on nova, release cycle of openstack is different from opnfv, how to allign?
Greg: These changes should be done in nova before we can use this API. In this group, only decide what to do, and then work in openstack?
Yuan Yue: suggest seperating into two parts, one for underlying messaging for communication. How to name server group? where to config the address of a VM?
Greg: will put more details in the doc
Yuan Yue: in the app layer, should we have an independent app run in VM for HA?
Greg: should be an app in the VM listening the report.
fuqiao: ask if it is possible to have a seperate service running in the host instead of working with the openstack nova team and promote these changes. Working with OpenStack will cost a long time and can hardly expect the result.
Greg: service runing on the host independent of OpenStack is possible. Will look into this.
The team agreed to review the doc again after the suggested changes made and an overview doc is provided. Greg will work on this in the following week and upload a new patch on gerrit for review.


Meeting on 13:00-14:00 UTC, Wednesday, Nov. 2nd


1, D release planning discussion

1) Gap analysis, try to engage  in the openstack community and promote the bps figured out during the C release.

2) HA API. Ian and Greg will work on uploading the API docs onto the OPNFV gerrit for review in the following week.

3) Bryan suggests to put a link of the HA testing on the wiki page.


2, Openstack summit sharing

Most of the workload currently in openstack are telecom issues. For the HA team, since we are focusing on single feature in each projects, it is necessary that we work with each different projects. Another suggestion would be to bring more resources so as to help promote these features.

Link to Bird of a Feather charts:

Meeting on 13:00-14:00 UTC, Wednesday, August 10th


Final call for comments on Gap Analysis Review : 

Yardstick team to join to provide an update and introduction to the new test framework


The Gap analysis review has been completed and successfully merged.  Thanks for all the feedback and suggestions.  Great team work to complete this for Colorado.

The Yardstick team members joined and gave a demo of the test results and the results on Jenkins

See here:

They also gave a great presentation on the Yardstick framework enhancements that they have been working on.  The presentation can be found here:

Thanks to the Yardstick team for joining our meeting and the great collaboration on the test cases.

Meeting on 13:00-14:00 UTC, Wednesday, July 13th


Review .RST file version of Gapa Analysis etherpad.



The focus of the meeting was to discuss the gap analysis.  
Gap 2: 
Needed significant rework.  The key issue is: if the nova compute service is down, but, VM’s are up.  What other tools are available to determine if the VM’s are actually available.  Tooz is an OpenStack project that was mentioned in the ether pad, it looks to be under active development. 
Gap 4: 
Update type to reliability/scalability. 
Action: Edgar to check status of blueprints mentioned. (Closed)
Gap 5: 
Need to mention that the listed blueprints aren’t sufficient, work is also needed in Nova/Neutron.  Multi-attach is a good start but more work is needed.
Action: Edgar to provide additional links to blueprints. (closed)
We agreed to remove the GAP on the ether pad that recommended a complete rearchitecture of the HA solution in OpenStack due to the large scope.  The recommendation is to find another vehicle for this proposal if the person who proposed it would like to continue the proposal.  We can add to the agenda for a future meeting.
General discussion at the end about how to increase momentum on HA work as it is critical to service providers.  

Meeting on 13:00-14:00 UTC, Wednesday, July 6th


Review open actions

Review offline comments on etherpad :

Close on which gaps to document and communicate.


End to end review of the etherpad -[ ] completed.  Final comments are due on the etherpad July 8th.

Ian to reformat etherpad data in rst and post for Gerrit review.  Live review during the next meeting.  

Yuan Yue confirmed status of many items with the Doctor team - Thanks! and the information is also captured in the Etherpad.


Meeting on 13:00-14:00 UTC, Wednesday, Jun. 29th


Final discussion on

- No comments posted on final review.  Looks like it is ready to go.

Discuss gap analysis etherpad :

- Review what is left and how to focus our efforts to close off this work item.


As it is vacation time, we had a small turn out today.  

We agreed to merge and this has now been done.

We kicked off our review of the etherpad on gap analysis.

We are asking for the team to do an offline review and we will have another meeting next week to roll in our feedback, close on the priorities and what we will document as gaps and focus areas.

Yuan Yue to check the status of the Doctor project work items with his colleagues who are working on Doctor.

Ian to check status of

We updated the etherpad with the notes from our discussion today.  Please use the etherpad to capture comments and ideas prior to our meeting on July 6th

Meeting on 13:00-14:00 UTC, Wednesday, Jun. 15th


1, yardstick HA test cases discussion


1, yardstick HA testcases


Recommend to add a step to show how to identify which HA proxy service is master.  So that the correct process is killed.
Add detail on pass criteria, discussion that the focus is to document the system behaviour.
Similar comments to TC006 - key is to identify which process to kill.  Public IP is what is going to be ping’d to measure the service impact.
Looks at impact of shutting down various Openstack services and determine how the system responds. Neutron, Keystone, Glance, Cinder and Swift are all covered.  
These are extensions to test cases executed during Bramaputra.
Test case priority: Planning to complete the 12 test cases prior to the end of Colorado cut offs.  
001,002,008-012 - first priority
006-007 - second level
003-005 - lowest priority and these test cases may not complete - Yardstick team working to implement HA team’s suggestions.
Lab availability - China Mobile will provide lab access - Yardstick team to contact Yang Jian to organize and plan the lab access. 

Meeting on 13:00-14:00 UTC, Wednesday, Jun. 1st


1, yardstick HA testcases

The HA team suggests the test case to 
1) make sure the netwrok interface is completely shutdown so that to triger the service failover
2) add another monitor to monitor the service on the failed controller node, so as to make sure it is isolated when the network fails.
The HA team also suggests another test case could be done in the future to test the HA of the network interface, e.g., when one of the nic fails, another bonding nic will take over the traffic and the service running on the host will not be influenced.

One thing to make sure is whether stressing the CPU to 100% will cause the total failure of the controller node. Otherwise there might be cases when some of the services on this specific node are still working. And it will be hard to tell whether it is the HA scheme that fails to triger the failover of service, or it is the fault that we inject not an efficient fault.

Same comments with tc003
1) make sure the disk I/O is completely failed to triger the service failover
2) add another monitor to make sure the failed node is isolated.

due to time limitation, the following cases will be discussed in the next call.

Meetings for High Availability Project

Meeting on 13:00-14:00 UTC, Wednesday, May. 18th


1, Deployment Guideline, Fuqiao will upload the new version by the end of this week. Please provide your comments and questions before that.

2, API work. Ian will upload the related slides and document and codes onto gerrit for further review.

3, Discussion about the Gap Analysis doc. Jingjing Niu from redhat was invited to join the meeting and provide help on the gap analysis and bps related. The 8 Gaps which were worked out by Yifei last year are discussed. Jingjing suggests we should look into the priority of these gap, and focus on one or two. Some follow up work is needed before we have further discussion.


Meeting on 13:00-14:00 UTC, Wednesday, May. 4th


Ian introduce the API between guest OS and host OS, which WR plan to opensource through OPNFV.

Usecases include:

1, Heartbeat

2, live migration rejection or preparation

3, VM peer state notification within server group

4, VM guest scaling of vCPU

Ian will upload related codes and docs onto the HA repository for further review.

Charts from presentation : 


1, work with installer team to see if we could include these API into one of the scenarios in the future release

2, API specification based on this is important

3, Will consider further opensource the e2e solution if the API draws the communities' attention.

Meeting on 13:00-14:00 UTC, Wednesday, Apr. 20th


Fuqiao present the deployment guideline doc. Got some comments from the team. Most questions are about the VNF HA and VM HA scheme. More explaination is necessary for this part.

Fuqiao will update the new version on the gerrit for further detail review.

Meeting on 13:00-14:00 UTC, Wednesday, Mar. 30th



1, Release C progress, Jira task update

Have update the jira to include the 3 tasks for colorado release.

HA-21 for deployment guideline, assigned to fuqiao

HA-23 for gap analysis, assigned to yifei

HA-24 for HA test cases, assigned to fuqiao

Next mailstone will be feature frozen. Time not confirmed yet.

Have discussion with senlin to see if we can include it in the deployment tools for colorado release. Senlin will be included in the Mitaka-RC1 release. It will be much easy to include it into opnfv once it is also included as an openstack release. ask yifei to see if this is possible. 

Bryan indicates that the model project is also intended to have discussion with the senlin project and see if they should also include senlin. Jointly discussion between the 3 teams might be helpful.

2, New committer vote

fuqiao will raise new committer vote for Ian and edgar in the mailing list.

3, Cross project meeting

Bryan is suggesting for cross project meeting instead of standalone project meetings.  It could be helpful for us to understand other projects and also know the progress. However we may need some extra internal meetings now and then to discuss about some project issues. We can also take full use of emails for internal discussion as well.

One possible style we can try is to have the cross project meeting once every month. Such meeting will include discussions about common subjects for multiple projects, testing and installations which are interested by multiple projects. Each project can have their standalone project meeting during the rest time of the month.


4, discussion on deployment doc and gap analysis doc

fuqiao made some progress on the etherpad. But have trouble with the wiki when they have some transfer work on wiki recently. Fuqiao suggests to continue the discussion on the etherpad.

gap analysis: yifei suggest we should begin with a few gaps, and try to promote back to openstack community. however we are lacking power at openstack.

Bryan suggests we could try to promote some demonstration, and can include some solutions to be built in in the future opnfv releases, so as to draw the attention of the openstack community as end users. We have already had two demos on opensaf and senlin last year during the OPNFV summit, which is quite successful. We can try to promote to include possible solutions into the colorado release.



Meeting on 13:00-14:00 UTC, Wednesday, Mar. 16th


1, Go through the outline of the HA deployment guideline doc.

Fuqiao will begin the documentation work on the etherpad.

1. Overview of High Available Deployment of NFV

(We would include all the possible modules that need to be high available, and give a general view of how to deploy each different module in a HA mode. However, not all of these HA schemes should be deployed in one system.)

1.1 Architecture of HA deployment

1.2 Hardware HA deployment topology

1.3 Software HA Framework

1.3.1 Controller services (Openstack services)

1.3.2 Hypervisor

1.3.3 Virtual Machine (VM)

1.3.4 Virtual Network Functions (VNF)

2. HA deployment guideline for OPNFV releases

2.1 HA deployment guideline for Arno

2.2 HA deployment guideline for Brahmaputra

2, Gap analysis doc

Yifei will update the doc, delete the gaps that has already been solved.

Meeting on 13:00-14:00 UTC, Wednesday, Mar. 2nd


1, Project Roadmap discussion

The following bullets are the plan for the 3rd-4th releases. We are planning to finish the first three bullet for the C release, and will have a start of the 4th one. We are expected to complete the HA API by the 4th release.

  • HA deployment for OPNFV: Already begin

1, continue the work, try to put forward a complete deployment guide by Release C.

2, This work will have to upgrade with the OPNFV releases as well.

3, Yuanyue suggests we should discuss about HA deployment with different deployment tools

  • Testing cases: Already begin

1, extend the test cases.

2, finish testing on other deployment tools

3, upgrade the user interface

  • Gap analysis: Already begin

1, finish the work on Gap Analysis of Openstack

  • HA API: Dependent on the work of ETSI NFV and OPNFV development

1, Work closely with other HA development teams, including OpenSAF and Senlin in Openstack. Try to unify the API.

2, Ian had some suggestions about begin with guest host API

3, ETSI work might need to wait for a while

4, Yifei suggest to work with movie team about VIM NB

Meeting on 13:00-14:00 UTC, Wednesday, Feb. 17th


1, Jonas introduces the OpenSAF project and their plan in the OPNFV. He addresses that the requirement doc for the HA project has provide the OpenSAF with some vision for how to develop the architecture to fit into NFV scenarios.

2, Jonas explained that OpenSAF will try to provide the following outputs in OPNFV

1) HA for VNF

2) HA as a service

3, Fuqiao introduce the HA project and the two demos during the last OPNFV summit.

4, The two teams reach agreements to cooperate in the following aspects:

1) Collaboratively provide HA schemes to OPNFV platforms (possible demos, codes to go into releases, and etc.)

2) Deployment guideline based on OpenSAF

3) Testing on OpenSAF solutions

4) HA API definition

5, the OpenSAF team would like to have one person to join the HA project meeting regularly, so as to know the progress of the two teams. And the two teams will have joint meetings about topics related with both of the projects.

Meeting on 13:00-14:00 UTC, Wednesday, Jan. 6th


1, HA testcases results

Yimin shares with us the test results of the HA tests in the China Mobile OPNFV testlab. The configuration parameters need more clarification. Will discuss these later in the mailing list

Ana explains their plan for these test results for the B release. Will have a document explain the test results. The HA team will review these test results as well, and make it as a joint B release output for the two teams.

More work need to be developed if we want to have a whole set of test of all the service of openstack. But due to the limited time before the deadline of B release, Ana suggests we can only finish these current tests and output the test results for the B release. We will continue the work later after the B release deadline. The China Mobile OPNFV Testlab can continiously provide the HA test infrastructure and will help with the testing.

The Yardstick test framework also will join the devotail project, and is plan to be one of the compliance tests.

2, state syncronization for gerrit

The deadline for B release is Feb. 2nd.

1) the general issue section and the multisite section have already been updated, need review from the whole team.

2) network node section:

improved a lot, had a few more questions need to be clarified. Yifei has updated new patch yesterday. Need review from the whole team.

3) Storage: Great work now. A few more questions in the review. Fuqiao will help abandon the old patch.

4) scalability: the team has a short discussion on whether we should include scalability issue in our scenario analysis doc. Edgar points out that the scalability issue is an availability problem, but not a HA problem. If we would like to narrow our scope of the project, maybe we should not include this section. But where should this issue be covered?

Ian suggests Yifei can first add a few more explaination about why scalability is an availability issue. Ian will provide comments on the gerrit about this. The other team members please also provide your ideas on the gerrit and we will decide whether we would like to include this section.

Meeting on 13:00-14:00 UTC, Wednesday, Dec. 16th


1, Committer list modification progress

fuqiao is currently working on the committer list and try to contact with those inactive committers. Most of them can be reached and have experessed understanding of this. Will organize vote for new committers after this process finishes.

2, Scenario doc discussion

1)multisite scenario-fuqiao

Maria has pointed out a great point, that in the multisite scenraio, we don't need to consider failure of VNFs. therefore we only need to work on the failure of infrastructures. However, it is also difficult to try to avoid influencing the performance and availability of the VNFs while providing solution for the HA of mulstisite scenario. For example, when the VM failed, the VNFs should failover to the other standby in the same site. but when the whole site fail, it should failover to the standby at the other site.

The team has reached agreements that we should cover the scenarios first, but should leave out the details of solution. Since we don't have successful solution now.

Edgar suggest to also take the maintanance usecase as an exampel as well.

2)scaling scenario-yifei

Currently the catogories is quite confusing. We also need to add more explaination about why scaling is related with HA, whether we have extra requirement for HA in the scaling scenario?

3)network node availability scenario-yifei

can we have a big picture of the HA of the network node first? For example, how to deploy HA for the network node. Maria has listed some suggested items on the gerrit.

4)storage scenario-edgar

Edgar introduced the basic structure, and he will reupload a new version to make the structure more clear.

3, HA testcase review

Yimin from the yardstick team has proposed a new template for the test case discription. May need review from our team.

Meeting on 13:00-14:00 UTC, Wednesday, Dec. 2nd


1, joint meeting with the Yardstick team for HA test cases discussion

Patric from the Yardstick team introduces the HA test cases. Lots of questions are poped for the test cases.

The first testcases seems quite clear after the meeting


The HA team will review the remain cases before the next meeting, and discuss with the Yardstick team on the etherpad.

May need another joint meeting after some discussion on the etherpad.

Meeting on 13:00-14:00 UTC, Wednesday, Nov. 18th


1, Design summit wrap-up

demo 1:

the demo can be the HA scheme for stateless VNFs, and for those stateful VNFs that do not have HA capability themselves, this demo is the best we can do. But we still would better require the application to have its own HA capability, and once the application is capable of, the NFVI should be consistantly used by the application, e.g., if the application have the state snychronization, the NFVI should not do the FT replication.

demo 2:

opensaf is a good solution for HA of applications. the OpenSAF forum will join OPNFV. we will invite the team to join our project discussion and see if it can also integrated with the senlin.

testing pools

testcases from yardstick team
Maria has checked the testcases gerrit, and may need some help understanding what the testcases exactly are.

2, Scenario analysis doc.

The deadline for the release B doc is appoaching. Fuqiao proposed we all upload the current sessions of the scenario analysis doc onto gerrit for review. Fuqiao has already upload the two sessions. Yifei and Edgar will work on their sessions as well.

Fuqiao has also upload the pdf version of the already finished doc onto gerrit:



Meeting on 13:00-14:00 UTC, Wednesday, Oct. 28th


1, OpenSAF under OPNFV

OpenSAF forum decided to close and join OPNFV as one of the project. There are some discussions in the TSC call, but not confirmed yet. Fuqiao will reaching out to the OpenSAF team and see if we could cooperate. We see that OpenSAF can be an important cluster service manager and HA deployment for VNFs and SDN controller.

2, Agenda for HA project breakout OPNFV Design Summit

1)short introduction of HA project

2) HA demo from China Mobile, Xinhui and Qiming

3) introduction of OpenSAF demo from Maria

4) Testcases of HA – peng hao from huawei

Time and location will be confirmed later

Fuqiao will not be at the summit this time. Maria will help hosting the HA project breakout.

3, Preparing for the "Introduction of HA project" on the Design Summit

Fuqiao will work on the slides

Meeting on 13:00-14:00 UTC, Wednesday, Oct. 14th


1, OPNFV Design Summit HA project session

The HA project will have a session during the design summit.

Possible agenda:

1) introduction of the HA project and our roadmap

2) HA demo from IBM and VMware

3) demo from Ericsson, Maria will share the results

4) discussion with the yardstick project, will talk with Ana (Not confirmed yet)

5) discussion with multisite project about geo redundancy usecases (Not confirmed yet)

Those who are planning to attend the HA session in the design summit can put their names on

We will also present our demo in the demo theater and on China Mobile booth:

2, committer remove process

The TSC is discussing about update the committer step down process for the project. Inactive committers will step down.
We will work on this once the process is approved by the TSC.

3, requirement doc progress

Section 3 has been on the gerrit for a while. Will be merged if no other comments received by the end of this week.

Section 5 is modifyed and a new patch is on the gerrit. ready for review.

Edgar suggest we should also have gap analysis doc for each different upstream, so as to put the requirements we work out into practice. The current gap analysis doc for openstack can be a good start.

Maria also suggest we should recheck requirement doc to see if it capture the requirements mentioned in the usecase analysis doc. We will merge all the sections of the requirement doc and do the recheck afterwards.

4, Scenario doc discussion

The fist part is on the gerrit for several weeks. Will be merged to the gerrit if no other comments proposed by the end of this week.

Maria suggests we should have seperate sections including the complex scenarios we currently work on. We will then build a complete doc for release after these docs are finished seperately.

Complex scenario.

multisite scenario: we had some discussion about geo redundancy scenario in multisite project. And Howard suggests, for the HA project, we will only focus on option 3 in the geo redundancy scenario, which will meet the HA requirement.

[action]Fuqiao will modify the draft on the etherpad.

storage part: Currently Edgar captured most of the HA schemes for the hardware layer. Maria suggests we should also take the NFVI and openstack into consideration. Here is the gerrit link of the gap analysis doc.

Meeting on 13:00-14:00 UTC, Wednesday, Sep. 23st


1, release B plan
we are moving to milestone C, building 2 jira story, one for requirment doc, assign to fuqiao, the other for scenario doc, assign to Maria

release doc close at Jan. 5th

release B on Feb. 2nd

template for the release has merged to the gerrit.

2, project life cycle changes

delete all the catogory def., delete top level project.

we may put more effort on deployment and testing after B release.

3, Vaccine project

Fuqiao gave some updates on the discussion with the Vaccine project. That project decided to postpone the project proposal. Part of the group join HA project testcase definition, part join yardstick work

4, requirement doc

only fuqiao and yuanyue's session under review.

Asking for review of section 3 by the end of this week.

Fuqiao will try to contact Yuan Yue. Or will merge section 5 and continue modify that section on a new thread.

5, scenario doc

ask for review for the basic usecases

plan for new sections:

1)network node HA scenario analysis – Yifei

2)storage node HA scenario analysis – Edgar

3)network failover – fuqiao

4) multisite usecases – fuqiao

5)scaling scenarios – yifei

Fuqiao will build a etherpad page so that we can input some draft in these section and have further discussion.

6 next ETSI meeting, Edgar will be there and will have another joint meeting with the REL WG

Meeting on 13:00-14:00 UTC, Wednesday, Sep. 10th


1, The multisite team is currently working on multisite usecase doc, and is also planning on including this in release B.

2, they currently have their usecase 1 and 3 on the gerrit. usecase 2 will be on gerrit by the end of this week. Usecase 2 and 3 is quite related to HA as well.

3, The multisite project will focus on requirements on openstack in these usecases. The HA project should look into these usecases and if there is any requirements on the other components, i.e., VNF, VNFM and etc., it should be addressed in the HA project.

4, Actions for the HA project:


The HA team will look into the related usecases in multisite and provide comments.
(will be in the gerrit by the end of this week)


Maria will extend the usecase doc including the multisite usecase. The usecases will refer to the work of the multisite group.


The multisite project will only focus on the requirement of Openstack. We shall look into these usecases and add extra requirements for the components apart from openstack.

5, For Usecase 3, the snapshot. The multisite team confirms they only focus on the requirement on openstack, therefore, they consider the application level HA, whether it is based on snapshot or application level replication, is out of the scope. They will only focus on letting openstack support all these activities application will take. Please provide more comments on the gerrit if you still have questions on this part.

Meeting on 13:00-14:00 UTC, Wednesday, Aug. 26st


1, Fuqiao introduces the scope of the testting work. And 2 simple cases was presented. Taking consideration of the current ARNO HA function, we decided to use the 1st testcase, which is HA testing for the Openstack services, as a beginning for the test.

2, Ana suggest we may need to know more details about the current ARNO deployment before we can have a detail procedure to do the testing. The Yardstick team will use the China Mobile lab and figure out this. May need help from the lab for reading the logs.

3, The two teams will cooperate in developing a full set of test cases and test tools for HA tesing of OPNFV. And hopefully we can do such testing on each following releases so that we can give instructions on what HA functionalities is included in the current release and what should be added afterwards.

Meeting on 13:00-14:00 UTC, Wednesday, Aug. 19st


1, Fuqiao introduces the presentation prepared for the Nov. OPNFV summit. The speakers of presentation will also include Qiming Teng from IBM and Xinhui Li from VMware.

Maria mentioned she would also like to share some demos about providing HA using OpenSAF. She will check out if this demo could be open to the public and we will see how to combile these two demos.

2, The whole team goes through the process of Gerrit.

Session 3 need to be split into two subsession, one is traditional hypervisor, the other is the middleware added for Carrier Grade NFV.

Session 4 needs modification by Ian.

Session 5 needs some more work from Yuan Yue.

Usecase doc is still worked on. Expect some update recently.

3, Fu Qiao talk about the cooperation with Yardstick about HA testcases and test tool development. We will have a joint meeting next week with the Yardstick team. We will suggest some test cases as a start, and the Yardstick team will help developing some test tools.

Fu Qiao suggest this work could continiously expand, so that we can finnaly have a full set of testcases of HA and also a full set of test tools. For each release of OPNFV, we can do these tests and refer to the whole community what cases has passed for this release and what else can be done.

Fu Qiao will work out some basic test cases and send out to the mailing list. Need discussion within the mailing list to see which case is better to be a start for such cooperation and development.

Meeting on 13:00-14:00 UTC, Wednesday, Aug. 19st


1, gerrit update.

Section 2, Section 3, Section 4 have got no comments in the passing weeks. Call for the whole team to give final comments. If no further comments by the end of this week, fuqiao will merge these 3 section.

Section 6 received some comments from Lijun about the fomat of rst file. Larry will help looking into this. Fuqiao will reach out to Lijun to check out what is missing for the whole project.

Section 5 receives some comments, need modification by Yuan Yue.

2, OPNFV Hackfest update

Fu Qiao gave some update about the REL WG joint meeting last week. The REL WG shows some interest in our work, and hope we can also refer to their work in the scenario analysis doc and requirement doc. They also gave some suggestions about the HA API work. They suggest we refer to IFA 005 and IFA 006 for the or-vi and vnfm-vi interfaces. The hardware interface is out of scope of ETSI NFV. And there is no work on the vnf-nfvi interface currently. They suggest if we have requirement for this interface, we can work it out and output to ETSI as well.

REL WG suggest to find a specific person both interested in the REL WG and the HA project to build the connection between these two projects. No one raise hand on the meeting to be this connection from our team though. Will reach out to the team.

3, Scenario analysis doc

Maria has prepared a draft for this doc and has updated it to the gerrit. Haven't got much comments on the call. The structure seems great and Fuqiao suggest to use this structure and finish the draft by including all the usecases already discussed in the slides. Further discussion is needed to include more usecases (multisite and scaling)

Meeting on 13:00-14:00 UTC, Wednesday, Jul. 29st


Meeting on 13:00-14:00 UTC, Wednesday, Jul. 22st


1, Fu Qiao introduced the plan for the 2nd and future releases.

2, The agreement is reached that we will propose to release the requirment doc and the scenario analysis doc for the 2nd release.

3, Maria will prepare a draft for the scenario doc and bring to the team for discussion

4, Fuqiao will try to talk with the ETSI NFV REL WG for HA API work.

5, There are suggestions that the release of our project should also include interfaces with the other projects (e.g. Multisite, Escalator, Doctor, and etc.). Maria will try to address the multisite scenario in the scenario draft and point it to the multisite project for further info (may need further discussion with the multisite project)

6, Yuan Yue suggest our gap analysis should not only cover openstack, but also cover ODL, pacemaker, and even OPNFV itself.

Meeting on 13:00-14:00 UTC, Wednesday, Jul. 15st


Meeting on 13:00-14:00 UTC, Wednesday, Jul. 8st


1, Qiming Teng introduces his work in Openstack. Include approach to realize HA in Openstack. He is currently working on the Senlin project, which provides cluster service. In the Senlin Project, HA can act as a policy.

Yuan Yue asked if this can provide HA for the VNF. Qiming answered that the Senlin project is only providing the cluster service, you can use it at the host level, the VM level and the VNF level. But currently it can not provide HA for stateful and stateless VNFs.

Fu Qiao asked if this still have problem of scalability since it use pacemaker as the cluster engine. Qiming explained yes, but they are trying to solve this by using cluster of clusters.

We may seek for future cooperation with the Senlin Project.

2, Gerrit update.

Session 1 is merged. Thank you for Ian.

Other sessions still have comments not solved. Need modification.

Meeting on 13:00-14:00 UTC, Wednesday, Jul. 1st


Meeting on 13:00-14:00 UTC, Wednesday, Jun. 24th


1, Gerrit Updates.

Section 1 and Section 6 seems to be matured. The review will be closed by the end of this week and we will summit the final patch at the beginning of next week.

Other sections need more comments and modifications.

2, Deployment framework

Fu Qiao shared a few slides about the deployment guideline. Gaps for the current Openstack and OPNFV platforms can be found. Further discussion and testing should be done before this deployment guideline is ready.

Meeting on 13:00-14:00 UTC, Wednesday, Jun. 17th


Meeting on 13:00-14:00 UTC, Wednesday, Jun. 10


1, Gerrit Update.

Section 1-4, section 6 have been uploaded to Gerrit.

[action] Call for the whole team to review the sections. Hope to finish the review by Friday, and finish the modification before next meeting.

Yuan Yue is still working on Section 5.

[action]Hope to upload to Gerrit by Friday.

Yifei has uploaded the Gap Analysis.

[action] Call for the whole team to review the analysis. Hope to finish the review before next meeting.

2, Future working item discussion

Fu Qiao has suggested several future working items for this project. Several team members gave comments and suggestions. The main thoughts are captured as follows.

1) API definition. One of the working items included in our project proposal.

There are suggestions that this work should be coordinated with the ETSI NFV work. The ETSI NFV will release its requirement doc by July. We shall arrange a discussion to make sure our requirement doc be noticed by the ETSI NFV WG. And further working on the the API definition in a collaborative way.

2) HA related test cases and specification development.

This working item is suggested by the TSC. We may need testing experties to join this work. In the meantime, what our requirement project should do is to generate test cases and specifications, while the testing project will help excuting the tests.

3) Reference deployment architecture for High Available OPNFV platform.

We shall develop deployable architectures for a high available OPNFV platform. Such architecture may include, for example, VNF should have its own HA at the VNF level, Storage HA architecture, and etc..Both Maria and Georg's work can be a good start. This architecture will provide guidance for people to deploy a high available end to end NFV environment.

Meeting on 13:00-14:00 UTC, Wednesday, Jun. 3rd


1, Gerrit progress

[action] all requirement doc sections will be upload to gerrit for review by the end of this week.
[action] Need review from the majority of the team

2, Storage section

Will be merged to multiple sections instead of being a standalone section.

[action] Georg will add comments of adding storage requirements to each related sections

Meeting on 13:00-14:00 UTC, Wednesday, May. 27th


1, fuqiao summirizes the HA session during the openstack summit.

2, Fuqiao introduces the discussion about cooperation between HA project and ETSI NFV project during the ETSI NFV meeting in Sanya. REL WG in ETSI NFV suggest we can have regular communication between the two projects. Some one from the the two projects can act as interface.

3, Ian and Maria give updates about the Git/Gerrit

[action] Fuqiao will look into creating jira ticket for the project output, and will also working on builing the derictory of the master repo
[action] Fuqiao will work on delivering section 3 for git review
[action] Ian will work on delivering section 4 for git review
[action] Yuan Yue will work on delivering section 5 for git review

4, Georg gives a summary of the storage HA proposal during the Openstack summit. will post the slides to the wiki page and the mailing list for further discussion

[action] Georg will try to add the requirement of the storage HA in the proposal into section 7 in the requirement doc.

Meeting on 13:00-14:00 UTC, Wednesday, May. 13th




1, HA scenario discussion. Conclusion may inlcude as follows:

1) Redundancy is required at the VNF level so as to meet the SAL recovery time of Carrier-Grade HA requirement.


2) VNF failure detection and failover can be managed by the VNF or the VNFM. There could be multiple ways to realized the HA schema in the VNF level.

2, slides for the hackfest is discussed. Work about the scenarios should be added into.

3, requirement doc. Session 3,4,5,6 can be delivered to the Git. Storage session needs to be discussed more.



Meeting on 13:00-14:00 UTC, Wednesday, May. 6th




1. Git/Gerrit update

[action] Fu Qiao will contact Aric to make sure adding all the committers and contributors to the Git repo
[action] Fu Qiao will look into the committer list to see if those unactive committers want to quit this project
[action] Fu Qiao will organize vote for new committers for the active contributors after the committer list is re-organized.

2. Vancouver meeting material discussion – Fu Qiao

[action] Fu Qiao will coordinate the input and send out the update version this week
[action] Yuan Yue will work on 1 page for the Monasca BP

3. Requirement doc update – Fu Qiao

[action] Fu Qiao will post to the mailing list to see whether the team thinks the other sections are mature

Meeting on 13:00-14:00 UTC, Wednesday, Apr. 29th


1. Maria mentioned about the cooperation with doctor and esculator project and ETSI HA WG.

[action] Maria will help send the email of the contacter from ETSI
[action] Fuqiao will follow up the cooperation discussion and try to arrange a f2f meeting before or during next month's ETSI meeting.

2. Go through the requirement doc

[action] section 3: provide specific requirement for the nova agent
[action] section 4: see if we have any specific requirement for vLB and vFW in this section
[action] section 5 and 6: ask for more comments from the team. Yuan Yue and Maria will try to look into the comments and make modification.

3. Sean share the HA work in openstack

There are no native HA schema in openstack. It's the architecture operators deploy that provides HA protection for openstack. Most API for openstack use active/active, and L4-L8 LB. Neotron API can not provide active/active HA architecture. Data storage traditionally support HA. Different hypervisor also support different features (example of live migration). Architecture of storage also have influence on HA and latency. Gap of openstack is mainly on the network side (ovs). We can look into the HA architecure of Fuel and Formane to have more general idea of HA for openstack.

[action] Sean will help review the Gap analysis doc, and provide comments this week
[action] Sean will help add a section about the HA requirement for storage in the requirement doc

4. Gerrit delivery

[action] Ian will help getting started with the Gerrit. May need some offline telephone call early next week

5. Vanconver Meetup

[action] Fuqiao will follow up with Ray to see we can have a timeslot on the OPNFV day


Meeting on Wednesday, Apr. 22nd


  • Ian presented the introductory section of the requirement document

o Maria suggested to look at the Doctor project’s requirement document, also that compared to Doctor the HA project has a bigger scope touching different layers and planes in the system, which would be good to point out

o Yuri suggested to add a figure that shows this bigger scope and indicate what are the different targets with respect to these different subsystems

o [action] Ian will try to integrate these comments
  • What kind of availability we want to support for VMs

o Agreement that VIM should control it ()

o Yuri said VM recovery at this level cannot satisfy stringent time requirements as it is today and that requires VNF solution

o Hui Deng suggested that it’s a feature nice to have, so we should look at possibilities

  • Are we ready to transfer the etherpad document to gerrit

o Ian suggested to review the etherpad document and decide

o Ian has experience with gerrit and can help with the transition

o Conclusion is that the sections 1.2-1.4 and 2 are in good shape to transfer others need more discussion

? We seem to be OK with the time requirements as they are currently

o Section 3 needs a bit more review/elaboration

? question: Is the nova agent on the compute node part of the VIM?

o Section 4

? Instead of addressing the load balancer and other NF individually we should address the HA requirement of NFs
provided in the infrastructure (e.g. load balancer, firewall)

Meeting on Wednesday, Apr. 8st




1. HA Blueprint discussion--Zhipeng, Joe


[Ian] since the multisite project is not created yet, this bp may be different from the architecture the multisite team eventially works out.

[Maria] Questions about the redundancy scheme for the token for different site.

[yuanyue] This bp may have extra requirement for nova and neotron to configure a secendary keystone server.

[yuanyue] Does this bp have extra requirement for the network between the different site. (chaoyi: may have some requirement for the latency)

[Ryota] since this bp is related with the keystonemiddle ware, is keystone the right place to propose?

2. HA requirement doc discussion--fuqiao


the whole team please review the requirement document and input comments and questions on the etherpad page.


Ian and Yuanyue will go through the Virtual infrastructure part and the VIM part and check out if the modification is ok. Please make modification if you find other requirements in other parts which may have relationship with your parts.


Ian will work on the frame work of the HA


fuqiao will work on the definition part


Yifei will go through the requirement doc and see if there is other gap from our requirements

Meeting on Wednesday, Mar. 25th


1, Virtual infrustucture Part-Ian

comments and modifications are put into the etherpad page.

[action]Ian will work on the modification
[action]Ian will try to propose a framework for the HA schemes.

2, gap analysis

1) [action]check out scedular part is in nova or not

2) pacemaker-servicegroup-driver

pacemaker in the compute node may not scale well

Ted:does pacemaker need a driver?

[action]may need further discussion offline

The group reached an agreement that the gap exists

[action]Yifei will modify this part to address the gap, and put the packmaker BP as one of the solutions.

Meeting on Wednesday, Mar. 17th


1, Ted: adding HPI and openHPI into the architecture work.

[action] Ted will help Yuan Yue on this.

2, For now, the VIM HA part will only address the HA scheme for VIM. And we will go through the whole document after we finish each part and see what is missing, and also to coordinate each part.

3, Ted: should not address floating IP in the requirement document. Should just point out high lever requirements.

[action] Yue will modify this.

4, Ryota: neotron architecture now is not good, should have seperate API and process

Yue: that is why we define 2 kinds of nodes, stateless and stateful, and define different policy for these two.

5, Ted: who to report to the alarm, what component to monitor the compute node?

Yue: Hardware controller will monitor the hardware for the compute node.

Ted: suggest to also add compute alarm requirements (already done in other parts)

6, fuqiao: who will be in charge of control node HA:

Yue: heatbeats between control nodes. only the software solution.

7, Maria: alarm also should be logged

[action] Yue will add this bullet.

8, Maria: should provide some criteria on the VIM HA, such as five nines, or the outage time

Yue: how to accomplish these numbers?

Ian: suggest to have five nines, since VIM is very important.

Maria: suggest at lease four nines

[action] we can have more discussion on the mailing list, or will continue this topic next time.
[action] Yue will update the etherpad page of the ViM HA part

Meeting on Wednesday, Mar. 4th


High Availability Project meeting

HA project weekly meeting

Wed, Mar 4, 10:00 PM Beijing, (6:00am, PST)


Please join my meeting from your computer, tablet or smartphone.


You can also dial in using your phone.


United States


(Long distance):
+1 (646) 749-3131


Access Code: 470-792-893


More phone numbers:




1, BP for Openstack L release

2, requirement doc update

• VIM part: Yuan Yue

3, Gap analysis of Openstack (Yifei Xue)

4, arrangement of our next meeting

Meeting on Thursday, Feb. 26th


High Availability Project meeting

HA project weekly meeting
Thu, Feb 26, 10:00 PM Beijing time (6:00am PST)


Please join my meeting from your computer, tablet or smartphone.


You can also dial in using your phone.


United States


(Long distance):
+1 (646) 749-3122


Access Code: 318-543-973


More phone numbers:


1, update of Prague Hackfest

2, BP for Openstack L release

3, requirement doc update

• VNF part: Maria

• VIM part: Yuan Yue

4, Gap analysis of Openstack (Yifei Xue)

5, arrangement of our next meeting

Meeting on Wednesday, Feb. 11th


High Availability Project meeting

Wed, Feb 11, 10:00 PM Beijing (6:00am PST)

• Please join my meeting from your computer, tablet or smartphone.

• You can also dial in using your phone.

United States (Long distance): +1 (646) 749-3122

Access Code: 767-529-453

More phone numbers:


1, Contact person for the project

2, Etherpad page for the requirement document. Please feel free to add your comments, suggestions and inputs directly into this page.

3, requirement doc update

• software part: Fu Qiao

• VNF part: Maria

4, arrangement of our next meeting

Meeting on Wednesday, Feb. 4th


Wed, Feb 4, 10:00 PM, Beijing time. (6:00am, PST.)

• Please join my meeting from your computer, tablet or smartphone.

• You can also dial in using your phone.

United States (Long distance): +1 (571) 317-3129

Canada (Long distance): +1 (647) 497-9353

Access Code: 631-890-413

More phone numbers:


1, requirement doc update

• Hardware part: Fu Qiao

• software part: Fu Qiao

• Virtual infrastructure part: Ian

• Vim part: Yuan Yue

• VNF part: Maria

2, discussion about session in Prague.

• Possible topics:

What do requirement docs need to include? Usecases or bullet of requirements?

3, arrangement of our next meeting

  • No labels