I am seeing that LACP frames are sent even when the master 'team' device is down, but the member ports are UP. Is this a bug or expected behavior?
Thanks Sridhar
Wed, Aug 26, 2015 at 11:19:38PM CEST, sridhar.samudrala@intel.com wrote:
I am seeing that LACP frames are sent even when the master 'team' device is down, but the member ports are UP. Is this a bug or expected behavior?
Hmm. I think that if team master is down, they should not be send.
What we need is to look for master admin_state event in teamd_lacp.c and enable/disable lacpdu frames sending.
Thanks Sridhar
On 8/27/2015 12:34 AM, Jiri Pirko wrote:
Wed, Aug 26, 2015 at 11:19:38PM CEST, sridhar.samudrala@intel.com wrote:
I am seeing that LACP frames are sent even when the master 'team' device is down, but the member ports are UP. Is this a bug or expected behavior?
Hmm. I think that if team master is down, they should not be send.
What we need is to look for master admin_state event in teamd_lacp.c and enable/disable lacpdu frames sending.
Is it OK to check for master admin_state before calling lacpdu_send()? OR Are you suggesting tracking admin_state in team_ifinfo and watch for changes in admin_state in teamd?
Thanks Sridhar
Fri, Aug 28, 2015 at 07:14:55PM CEST, sridhar.samudrala@intel.com wrote:
On 8/27/2015 12:34 AM, Jiri Pirko wrote:
Wed, Aug 26, 2015 at 11:19:38PM CEST, sridhar.samudrala@intel.com wrote:
I am seeing that LACP frames are sent even when the master 'team' device is down, but the member ports are UP. Is this a bug or expected behavior?
Hmm. I think that if team master is down, they should not be send.
What we need is to look for master admin_state event in teamd_lacp.c and enable/disable lacpdu frames sending.
Is it OK to check for master admin_state before calling lacpdu_send()? OR Are you suggesting tracking admin_state in team_ifinfo and watch for changes in admin_state in teamd?
The second approach looks much nicer to me.
Thanks Sridhar
libteam@lists.fedorahosted.org