Hi Eric,

 

I retested it. And here is all the information you requested. I uploaded them to my google drive. Below is the link.

 

https://drive.google.com/open?id=1LiCgPz5iS3SApb0Em8ZyA56lryAU-3Vm

 

I conducted two tests. For case 1, the system can recover. For case 2, it can’t.

Case 1: pull the active controller management cable for longer than 30s, and then reinsert it

Case 2: pull the active controller management cable for less than 30s, and then reinsert it

 

For case 2, since controller-1 was shown as “offline” on controller-0. “collect all” can’t get the information of controller-1. So I copied the whole folder “/var/log” of controller-1. It is included in the shared zip package. If you need more information, let me know. Thanks for your help again!

 

Thanks.

Yi

From: Wang, Yi C
Sent: Tuesday, July 2, 2019 9:21 PM
To: MacDonald, Eric <Eric.MacDonald@windriver.com>
Cc: starlingx-discuss@lists.starlingx.io
Subject: RE: [Starlingx-discuss] a question about starlingx error handling behavior

 

Hi Eric,

 

Thank you, Eric! I will collect all the information and get back to you soon.

 

I confirm that I physically pulled the cable.

 

Thanks.

Yi

From: MacDonald, Eric [mailto:Eric.MacDonald@windriver.com]
Sent: Tuesday, July 2, 2019 7:54 PM
To: Wang, Yi C <yi.c.wang@intel.com>
Cc: starlingx-discuss@lists.starlingx.io
Subject: Re: [Starlingx-discuss] a question about starlingx error handling behavior

 

Please perform both cases and for each

-         indicate what cables/interfaces were pulled

-         indicate what hosts the cables were pulled from

-         indicate approximate timestamp of when the cable was pulled

-         indicate approximate timestamp of when the cable was reinserted

-         Run ‘collect all’ and provide me access to the collect tarball

 

Also, just to be sure … please confirm that you are physically pulling the cable and not just ifdowning the interface.

 

Eric.

 

From: Wang, Yi C [mailto:yi.c.wang@intel.com]
Sent: Sunday, June 30, 2019 9:45 PM
To: MacDonald, Eric
Cc: starlingx-discuss@lists.starlingx.io
Subject: [Starlingx-discuss] a question about starlingx error handling behavior
Importance: High

 

Hi Eric,

 

I am working on the LP 1815513. Based on my tests, if I unplug the cable of active controller for management network for a long time (for example, 30s), and then plug it, the whole system can recover after some reboots. But if I unplug the cable for a short time, and then plug it. The whole system can’t recover. I need to lock/unlock controllers manually to bring the system back. So my questions are:

 

1. Is the behavior acceptable? (recover the system by manual lock/unlock operations)

2. If the answer is no for #1, we need the system to recover automatically. I am not familiar with internal maintenance logics, could you give me some hints?

 

Thanks.

Yi