[Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Liu, ZhipengS zhipengs.liu at intel.com
Wed Jun 3 02:03:42 UTC 2020 

Hi Frank,

Thanks for your quick update!
Which build are you using to test this case? 
Since decoupling commits introduced several regressions (at least 2),  not propose to do this kind of stability test with latest build. 
BTW, do we have plan to revert them considering this stability risk?  Our Ussuri upgrade patches is waiting for it☹

Furthermore, we have not seen this test case that force reboot both controllers at the same time. Is it a new requirement?  If not , have we pass this case before, which build?  
I'd like to help on it with the pass build for comparative analysis. From my point , mariadb might not work if we reboot both controllers.

Thanks!
Zhipeng

-----Original Message-----
From: Miller, Frank <Frank.Miller at windriver.com> 
Sent: 2020年6月3日 8:55
To: Miller, Frank <Frank.Miller at windriver.com>; Liu, ZhipengS <zhipengs.liu at intel.com>; starlingx-discuss at lists.starlingx.io; Church, Robert <Robert.Church at windriver.com>
Subject: RE: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Zhipeng:

An update on our testing and analysis today.  We are able to reproduce the issue with OpenStack not recovering when we trigger a reboot of both AIO controllers at the same time.  This results in MariaDB and multiple other OpenStack pods in CrashLoopBackoff and openstack commands not working indefinitely after the controllers recover.  We'll create a launchpad tomorrow to track this issue.

Frank

-----Original Message-----
From: Miller, Frank <Frank.Miller at windriver.com>
Sent: Tuesday, June 02, 2020 12:25 PM
To: Liu, ZhipengS <zhipengs.liu at intel.com>; starlingx-discuss at lists.starlingx.io; Church, Robert <Robert.Church at windriver.com>
Subject: Re: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Thanks Zhipeng for the analysis.  What is challenging here is the multitude of issues.  

In our debug of openstack the past few days we are seeing the app fail completely.  After investigation this issue is a Day 1 containerd issue.  This is tracked in LP: https://bugs.launchpad.net/starlingx/+bug/1881353 

The issue you are seeing on a swact is a new and very recent issue tied to the decoupling commits that were merged late last week.  Bob is investigating and I expect he'll have a fix soon for that.

But the issues we are most concerned with are when we see mariadb crashing and not able to recover or with openstack services not working for longer periods of time.  We're attempting to isolate the sequence of events that trigger this.

Frank

-----Original Message-----
From: Liu, ZhipengS <zhipengs.liu at intel.com>
Sent: Tuesday, June 02, 2020 11:47 AM
To: Miller, Frank <Frank.Miller at windriver.com>; starlingx-discuss at lists.starlingx.io; Church, Robert <Robert.Church at windriver.com>
Subject: RE: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

For LP https://bugs.launchpad.net/starlingx/+bug/1881454
            Unable to unlock controller after swact and lock w/ openstack applied I also tested with daily build 20200516T080009Z. However, it could not be reproduced.
We should  fix this regression ASAP!

Thanks!
Zhipeng

-----Original Message-----
From: Liu, ZhipengS <zhipengs.liu at intel.com>
Sent: 2020年6月2日 16:48
To: Miller, Frank <Frank.Miller at windriver.com>; starlingx-discuss at lists.starlingx.io; Church, Robert <Robert.Church at windriver.com>
Subject: Re: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Hi Frank and all,

Update for issue 2.
I raised a new LP to track it.
https://bugs.launchpad.net/starlingx/+bug/1881722
Below is the time statistics. It seems reasonable. No obvious issue found. 
1) 3~4min for host restart and get ready.
2) 2~3min for mariadb terminating, initialization, get ready. (then configmap sync is ready)
3) 2min for ovs-db ready (reduce probe live/ready timer can improve a little, as it can retry quickly to connect ovs-vsctl: unix:/var/run/openvswitch/db.sock)
4) 1min for other pods ready, like neutron-ovs-agent which depends on ovs-db. ) Any comment?

For LP https://bugs.launchpad.net/starlingx/+bug/1881454
            Unable to unlock controller after swact and lock w/ openstack applied
   And  https://bugs.launchpad.net/starlingx/+bug/1881711
            system helm-override-show stx-openstack mariadb openstack crash  It seems related to openstack plugin decouple related patches. Should be a regression. 
 Please see our update in this 2 LPs for detail info.  @Bob, could you pls help further check it and your patches, thanks!

Thanks!
Zhipeng

-----Original Message-----
From: Liu, ZhipengS
Sent: 2020年6月1日 16:20
To: 'Miller, Frank' <Frank.Miller at windriver.com>; 'starlingx-discuss at lists.starlingx.io' <starlingx-discuss at lists.starlingx.io>; Jascanu, Nicolae <nicolae.jascanu at intel.com>
Subject: RE: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Hi Frank,

I also tested the issue 2 with latest daily build on duplex setup.
The conclusion is that the issue is there all the time. 
This issue might not be fixed soon, but should not block OpenStack upgrade, right? 

For 9 OpenStack patches below, I have removed all workflow-1, except the first patch and add depends-on all them.
https://review.opendev.org/#/q/topic:for_ussuri+(status:open)
Your review and comments are welcome!

As for issue 2, some detail info FYI.
It also needs to wait for around 10 min before all pods are ready again after reboot for master build.
It stuck on below 2 pods for 10 min. The same as the one I saw with my OpenStack upgrade engineering build. 
     neutron-ovs-agent-controller-0-937646f6-xxznw(depends openvswitch-db)
     openvswitch-db-8fxkw
Related key logs below.
  Warning  FailedMount  2m19s              kubelet, controller-1  MountVolume.SetUp failed for volume "openvswitch-db-token-7g4qk" : failed to sync secret cache: timed out waiting for the condition
  Warning  FailedMount  2m19s              kubelet, controller-1  MountVolume.SetUp failed for volume "openvswitch-bin" : failed to sync configmap cache: timed out waiting for the condition
  Warning  FailedMount  105s               kubelet, controller-1  MountVolume.SetUp failed for volume "openvswitch-db-token-7g4qk" : failed to sync secret cache: timed out waiting for the condition
  Warning  FailedMount  105s               kubelet, controller-1  MountVolume.SetUp failed for volume "openvswitch-bin" : failed to sync configmap cache: timed out waiting for the condition
  Warning  Unhealthy    30s                kubelet, controller-1  Liveness probe failed: ovs-vsctl: unix:/var/run/openvswitch/db.sock: database connection failed (Permission denied)
  Warning  Unhealthy    7s                 kubelet, controller-1  Readiness probe failed: ovs-vsctl: unix:/var/run/openvswitch/db.sock: database connection failed (Permission denied)

Is it the same stability issue as the one reported from your test team?  I can only see this issue after force rebooting.  What is our expected recovery time?
Your comment is appreciated! 

Thanks!
Zhipeng
-----Original Message-----
From: Liu, ZhipengS
Sent: 2020年5月29日 9:42
To: 'Miller, Frank' <Frank.Miller at windriver.com>; starlingx-discuss at lists.starlingx.io; Jascanu, Nicolae <nicolae.jascanu at intel.com>
Subject: RE: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Hi Frank,

Glad to see your quick reply!!
For OpenStack upgrade task, we have finished all test and get patches ready for more than 2 weeks, but no any review comments and feedback from your side.  What's the next step?

For issue # 2,  in community meeting notes,  I saw that you had some stability issue from WR local test team. But so far, I do not see any LP for the detail info. You should ask them to do that!  Right?

According to your concern, I tried to reproduce it with my build (cherry pick OpenStack upgrade patches)yesterday, and the original issue [1] was not seen any more, mariadb got ready quickly, no regression. 

[1] https://bugs.launchpad.net/starlingx/+bug/1855474

Thanks!
Zhipeng

-----Original Message-----
From: Miller, Frank <Frank.Miller at windriver.com>
Sent: 2020年5月29日 1:07
To: Liu, ZhipengS <zhipengs.liu at intel.com>; starlingx-discuss at lists.starlingx.io; Jascanu, Nicolae <nicolae.jascanu at intel.com>
Subject: RE: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Thanks Zhipeng.

Good to see progress on IPv6.
Waiting for 10 minutes for pods to recover isn't a good result.  Is there a LP open on this issue?  Which pods are not ready? What can you tell us about this 10 minute outage?

Frank

-----Original Message-----
From: Liu, ZhipengS <zhipengs.liu at intel.com>
Sent: Thursday, May 28, 2020 5:06 AM
To: Miller, Frank <Frank.Miller at windriver.com>; starlingx-discuss at lists.starlingx.io; Jascanu, Nicolae <nicolae.jascanu at intel.com>
Subject: RE: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Hi Frank,

Nicolae already added test case description. Thanks Nicolae!

I also did below test on AIO-DX virtual setup, exactly according to your mentioned steps.
No issue found, but just need to wait for around 10 min before all pods are ready again after reboot.

For ipv6 issue, I have submitted new patch for it since dynamic override for database config did not work.
 https://review.opendev.org/#/c/731461/
 https://review.opendev.org/#/c/731470/

Thanks!
Zhipeng

-----Original Message-----
From: Miller, Frank <Frank.Miller at windriver.com>
Sent: 2020年5月27日 22:43
To: Liu, ZhipengS <zhipengs.liu at intel.com>; starlingx-discuss at lists.starlingx.io; Jascanu, Nicolae <nicolae.jascanu at intel.com>
Subject: RE: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Zhipeng:

Thanks for the info.  You have provided the # of testcases but not what those testcase do.  Where can I find a description of what the OpenStack testcases do?

For the controller reset testcases I'd like to see the test result for the following:
Is openstack usable during the following scenarios on AIO-DX and on Standard configurations:
- Lock/unlock of standby controller
- reset (ie: reboot -f) of the standby controller
- reset (ie: reboot -f) of the active controller
- reapply of stx-openstack after the above scenarios

Frank

-----Original Message-----
From: Liu, ZhipengS <zhipengs.liu at intel.com>
Sent: Wednesday, May 27, 2020 9:15 AM
To: Miller, Frank <Frank.Miller at windriver.com>; starlingx-discuss at lists.starlingx.io; Jascanu, Nicolae <nicolae.jascanu at intel.com>
Subject: RE: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Hi Frank,

We have done below tests.
1) Sanity tests by Nicolae.
AIO - Simplex
Setup                                    04 TCs [PASS]
Provisioning                       01 TCs [PASS]
Sanity OpenStack             49 TCs [PASS]
Sanity Platform                 07 TCs [PASS]

TOTAL: [ 61 TCs ]

AIO - Duplex 
Setup                                    04 TCs [PASS]
Provisioning                       01 TCs [PASS]
Sanity OpenStack             52 TCs [PASS]
Sanity Platform                 07 TCs [PASS]

TOTAL: [ 64 TCs ]

Standard - Local Storage (2+2)
Setup                                    04 TCs [PASS]
Provisioning                       01 TCs [PASS]
Sanity OpenStack             52 TCs [PASS]
Sanity Platform                 08 TCs [PASS]

TOTAL: [ 65 TCs ]

Standard External - Dedicated Storage (2+2+2)
Setup                                    04 TCs [PASS]
Provisioning                       01 TCs [PASS]
Sanity OpenStack             52 TCs [PASS]
Sanity Platform                 09 TCs [PASS]

TOTAL: [ 66 TCs ]

2) NFV scenario test by me
    on duplex/multi standard virtual setup
          duplex bare metal setup
===== Setup =================================================================================================================================
2020-05-14 02:30:05.524  Create flavor small  ........................................ [OKAY]
2020-05-14 02:30:05.524  Create flavor small_ephemeral  .............................. [OKAY]
2020-05-14 02:30:05.524  Create flavor small_swap  ................................... [OKAY]
2020-05-14 02:30:05.524  Create flavor small_ephemeral_swap  ......................... [OKAY]
2020-05-14 02:30:05.524  Create flavor medium  ....................................... [OKAY]
2020-05-14 02:30:05.524  Create flavor medium_ephemeral  ............................. [OKAY]
2020-05-14 02:30:05.524  Create flavor medium_swap  .................................. [OKAY]
2020-05-14 02:30:05.524  Create flavor medium_ephemeral_swap  ........................ [OKAY]
2020-05-14 02:30:05.653  Create image cirros  ........................................ [OKAY]
2020-05-14 02:30:05.695  Create volume cirros  ....................................... [OKAY]
2020-05-14 02:30:05.695  Create volume cirros-ephemeral  ............................. [OKAY]
2020-05-14 02:30:05.695  Create volume cirros-swap  .................................. [OKAY]
2020-05-14 02:30:05.695  Create volume cirros-ephemeral-swap  ........................ [OKAY]
2020-05-14 02:30:05.695  Create volume empty_volume  ................................. [OKAY]
2020-05-14 02:30:05.786  Create network internal  .................................... [OKAY]
2020-05-14 02:30:06.158  Create network external  .................................... [OKAY]
2020-05-14 02:30:06.772  Create subnet internal  ..................................... [OKAY]
2020-05-14 02:30:07.661  Create subnet external  ..................................... [OKAY]
2020-05-14 02:30:08.553  Create instance cirros-1  ................................... [OKAY]
2020-05-14 02:30:29.918  Create instance cirros-ephemeral-1  ......................... [OKAY]
2020-05-14 02:30:43.160  Create instance cirros-swap-1  .............................. [OKAY]
2020-05-14 02:30:56.101  Create instance cirros-ephemeral-swap-1  .................... [OKAY]
2020-05-14 02:31:09.077  Create instance cirros-image-1  ............................. [OKAY]
2020-05-14 02:31:21.241  Create instance cirros-image-with-volumes-1  ................ [OKAY] =============================================================================================================================================
===== Test Iteration 0 (single-execution) ===================================================================================================
2020-05-14 02:33:04.172  Test Instance-Pause  ........................................ [OKAY]  (2020-05-14 02:33:18.078 Δ=0:00:12.870)
2020-05-14 02:33:35.073  Test Instance-Unpause  ...................................... [OKAY]  (2020-05-14 02:33:41.608 Δ=0:00:05.866)
2020-05-14 02:33:53.049  Test Instance-Suspend  ...................................... [OKAY]  (2020-05-14 02:33:59.546 Δ=0:00:05.792)
2020-05-14 02:34:11.103  Test Instance-Resume  ....................................... [OKAY]  (2020-05-14 02:34:17.756 Δ=0:00:05.937)
2020-05-14 02:34:29.269  Test Instance-Reboot (soft)  ................................ [OKAY]  (2020-05-14 02:36:45.923 Δ=0:02:15.748)
2020-05-14 02:37:02.160  Test Instance-Reboot (hard)  ................................ [OKAY]  (2020-05-14 02:37:14.504 Δ=0:00:11.704)
2020-05-14 02:37:30.673  Test Instance-Stop  ......................................... [OKAY]  (2020-05-14 02:38:44.543 Δ=0:01:13.220)
2020-05-14 02:39:00.481  Test Instance-Start  ........................................ [OKAY]  (2020-05-14 02:39:07.198 Δ=0:00:06.068)
2020-05-14 02:39:18.578  Test Instance-Live-Migrate  ................................. [OKAY]  (2020-05-14 02:39:41.692 Δ=0:00:22.306)
2020-05-14 02:39:57.927  Test Instance-Cold-Migrate  ................................. [OKAY]  (2020-05-14 02:41:22.720 Δ=0:01:24.179)
2020-05-14 02:41:38.995  Test Instance-Cold-Migrate-Confirm  ......................... [OKAY]  (2020-05-14 02:41:45.441 Δ=0:00:05.884)
2020-05-14 02:41:57.108  Test Instance-Cold-Migrate-Revert  .......................... [OKAY]  (2020-05-14 02:43:36.381 Δ=0:00:21.637)
2020-05-14 02:43:52.320  Test Instance-Resize  ....................................... [OKAY]  (2020-05-14 02:45:16.409 Δ=0:01:22.812)
2020-05-14 02:45:32.723  Test Instance-Resize-Confirm  ............................... [OKAY]  (2020-05-14 02:45:39.119 Δ=0:00:05.777)
2020-05-14 02:45:50.437  Test Instance-Resize-Revert  ................................ [OKAY]  (2020-05-14 02:47:30.175 Δ=0:00:21.748)
2020-05-14 02:47:46.230  Test Instance-Rebuild  ...................................... [OKAY]  (2020-05-14 02:48:59.762 Δ=0:01:12.980)
Total-Tests: 16     Execution-Time: 0:16:11.676

3) Another 2 test
    a) Using IPv6
         It can pass with workaround now.  I need one more fix for it.
         In my previous patch https://review.opendev.org/#/c/716524 (merged), I dynamically override below 
            config_override: |
                [mysqld]
                bind_address=::
         However, it did not work now. From log,  it shows error "OpenStack-Helm Mariadb - INFO - b'error: Found option without preceding group in config file: /etc/mysql/conf.d/20-override.cnf at line: 1'"
         I tried many methods, but could not remove the first line in 20-override.cnf
                mysql at mariadb-server-0:/etc/mysql/conf.d$ cat 20-override.cnf
                |-     
                [mysqld]
                bind_address=::
        I can only add it in manifest.yaml as a static override like below.
               values:
                  conf:
                      database:
                          config_override: |
                              [mysqld]
                              bind_address=::
            
    b) Reset of controllers and check status of OpenStack while a controller is rebooting.
         I have tested it and pass on simplex.
         For duplex, I have a setup issue in my side.
         @Jascanu, Nicolae  Could you help me on it for duplex test, if you have time today. Thanks!

Zhipeng



-----Original Message-----
From: Miller, Frank <Frank.Miller at windriver.com>
Sent: 2020年5月26日 21:13
To: Liu, ZhipengS <zhipengs.liu at intel.com>; starlingx-discuss at lists.starlingx.io
Subject: RE: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Zhipeng:

Can you publish the list of tests that have been run for openstack?

Also has openstack been tested for the following scenarios:
1) Using IPv6
2) Reset of controllers and check status of openstack while a controller is rebooting?  

Frank

-----Original Message-----
From: Liu, ZhipengS <zhipengs.liu at intel.com>
Sent: Monday, May 25, 2020 3:14 AM
To: starlingx-discuss at lists.starlingx.io
Subject: Re: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Hi all,

We have passed all sanity test on all setup. Thanks Nicolae!!
We also built out OpenStack service images from layered build environment.

Please help to review and push below patches to be merged, thanks!
https://review.opendev.org/#/q/topic:for_ussuri+(status:open+OR+status)

BRs
Zhipeng 

-----Original Message-----
From: Liu, ZhipengS
Sent: 2020年5月14日 16:49
To: 'Saul Wold' <sgw at linux.intel.com>; 'starlingx-discuss at lists.starlingx.io' <starlingx-discuss at lists.starlingx.io>
Subject: RE: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Hi all, 

Call for patch review again!
https://review.opendev.org/#/q/topic:for_ussuri+(status:open+OR+status)

Thanks!
Zhipeng

-----Original Message-----
From: Liu, ZhipengS
Sent: 2020年5月9日 8:38
To: Saul Wold <sgw at linux.intel.com>; starlingx-discuss at lists.starlingx.io
Subject: RE: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

Agree!

-----Original Message-----
From: Saul Wold <sgw at linux.intel.com>
Sent: 2020年5月9日 0:29
To: starlingx-discuss at lists.starlingx.io
Subject: Re: [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call for patch review!!

I would strengthen that to no changes until we get Green Sanity other than what's required to make them Green.

Full Stop!

Sau!


On 5/8/20 9:05 AM, Miller, Frank wrote:
> Until we can get sanity passing for several days in a row I strongly 
> suggest we do not allow any further changes into the load related to 
> OpenStack.  Folks can continue with reviews but let’s hold off 
> allowing merges related to a new OpenStack version.
> 
> Frank
> 
> *From:*Liu, ZhipengS <zhipengs.liu at intel.com>
> *Sent:* Friday, May 08, 2020 11:59 AM
> *To:* starlingx-discuss <starlingx-discuss at lists.starlingx.io>
> *Cc:* YU CHENGDE <yu.chengde at 99cloud.net>; Penney, Don 
> <Don.Penney at windriver.com>
> *Subject:* [Starlingx-discuss] [OpenStack Ussuri Upgrade Task] Call 
> for patch review!!
> 
> Hi all,
> 
> Please help to review OpenStack Ussuri upgrade patches.
> 
> Our target is to get all below patches merged by end of next week.
> 
> https://review.opendev.org/#/q/topic:for_ussuri+(status:open+OR+status
> :merged)
> 
> During OpenStack upgrade for StarlingX, we have to move python2.7 to
> python3.6 for OpenStack services as ussuri release only support python3.
> 
> We also rebased openstack-helm/helm-infra to latest version.
> 
> Engineering build test status.
> 
>  1. nfv_scenario_tests PASS on simplex/duplex/multi virtual setup.
>  2. nfv_scenario_tests PASS on simplex bare metal setup.
>  3. Sanity test is ongoing.   Duplex/standard virtual setup test PASS.
> 
> Thanks!
> 
> Zhipeng
> 
> 
> _______________________________________________
> Starlingx-discuss mailing list
> Starlingx-discuss at lists.starlingx.io
> http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss
> 

_______________________________________________
Starlingx-discuss mailing list
Starlingx-discuss at lists.starlingx.io
http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss
_______________________________________________
Starlingx-discuss mailing list
Starlingx-discuss at lists.starlingx.io
http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss
_______________________________________________
Starlingx-discuss mailing list
Starlingx-discuss at lists.starlingx.io
http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss
_______________________________________________
Starlingx-discuss mailing list
Starlingx-discuss at lists.starlingx.io
http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss

More information about the Starlingx-discuss mailing list