Sanju,

Could you point me to the lauchpad/gerrit of the code fix?

Thanks,

Edgar

From: Sanju Abraham
Date: Thursday, October 15, 2015 at 9:49 AM
To: Edgar Magana, 
"[email protected]<mailto:[email protected]>", 
"[email protected]<mailto:[email protected]>"
Subject: Re: [opencontrail-dev] [users] Meta-data issue under concurrency

Edgar,

The 502 issue that we saw in our testing yesterday is fixed in release R2.20. 
Contrail >= 2.20 should have the fix.

Thanks,
Sanju

From: Edgar Magana <[email protected]<mailto:[email protected]>>
Date: Thursday, October 15, 2015 at 9:34 AM
To: "[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>, 
"[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>
Subject: [opencontrail-dev] [users] Meta-data issue under concurrency

Users and Developers,

We are hitting a very weird and random issue in contrail. We believe that is 
due to something wrong on the metadata_proxy.cc code.
Our test basically sends 30 concurrent requests to the metadata service and 
some of the them failed with a 502 “Bad Getway” response from contrail code.

After running tcpdump on both the tap interface of the VM and the vhost0 on the 
hypervisor where that VM is being hosted, we noticed that the VM sends properly 
the request but when the request fails is because we got another request coming 
from the vhost0 interface, it looks like the vRouter is forwarding back the 
meta-data request to the VM client instead of forwarding it to the meta-data 
server.

We have this case mostly uncer concurrent calls and sometimes even with 10 
concurrent calls we will be able to hit it. We were wondering is someone has 
experience this issue before.

Ou contrail version is 2.11-56.el6 on CentOS 6.5

Thanks,

Edgar


_______________________________________________
Dev mailing list
[email protected]
http://lists.opencontrail.org/mailman/listinfo/dev_lists.opencontrail.org

Reply via email to