Establishing a Link Aggregation with libteamd

[Andy Johnson]

Hello, libteam.

When trying to establish a link aggregation with libteam, against an
A/D switch (which is configured to support LACP), is there a way to
know what is the status of the LACP session by some userspace tool
(apart from a sniffer),  or by any logs ?

Thanks,

Andy

[Jiri Pirko]

Fri, May 30, 2014 at 12:54:06PM CEST, [e-mail] wrote:
> Hello, libteam.
>
> When trying to establish a link aggregation with libteam, against an
> A/D switch (which is configured to support LACP), is there a way to
> know what is the status of the LACP session by some userspace tool
> (apart from a sniffer),  or by any logs ?

the best way to get info from teamd is using teamdctl util. For example:

# teamdctl team0 state
# teamdctl team0 state dump

man teamdctl for more details.

that should be enough to get the info you need. If you need anything
else, please tell, the output can be easily extended.

Jiri
> Thanks,
>
> Andy
> _______________________________________________
> libteam mailing list
> [e-mail]
> https://lists.fedorahosted.org/mailman/listinfo/libteam

[Andy Johnson]

Hello,

Thanks for your quick response.

Still there is something strange when running "teamdctl team0 state",
,which bewilders me. I hope you can advice.

I have a simple setup where I have a Linux server connected to Cisco
5000 ADswitch. I had configured LACP on two ports in the Linux server,
and configured LAG on the Cisco 5000 switch.
After running the teamd daemon with the LACP mode and running
"teamdctl team0 state", I get data which shows that the LACP session
is ok (so I understand, see below the full output).

However, also after I remove the LAG from the switch and restart the
daemon, I still get the same output from
"teamdctl team0 state", which seems strange to me. I made sure that the
LAG is not running on the switch, by "show port-channel database" (I
get empty response).

The LACP session should have not succeed in this case, so how come
"teamdctl team0 state" is the same ? Or let's ask this: How does
"teamdctl team0 state" should show that the LACP session was not
established ?

(or how "teamdctl team0 state dump" should be different ? )

Following here is the full log of "teamdctl team0 state":

teamdctl team0 state
setup:
  runner: lacp
ports:
  p5p1
    link watches:
      link summary: up
      instance[link_watch_0]:
        name: ethtool
        link: up
    runner:
      aggregator ID: 0
      selected: no
      state: defaulted
  p5p2
    link watches:
      link summary: up
      instance[link_watch_0]:
        name: ethtool
        link: up
    runner:
      aggregator ID: 0
      selected: no
      state: defaulted
runner:
  active: yes
  fast rate: yes

Regards,
Andy

On Fri, May 30, 2014 at 4:01 PM, Jiri Pirko  wrote:
> Fri, May 30, 2014 at 12:54:06PM CEST, [e-mail] wrote:
> > Hello, libteam.
> >
> > When trying to establish a link aggregation with libteam, against an
> > A/D switch (which is configured to support LACP), is there a way to
> > know what is the status of the LACP session by some userspace tool
> > (apart from a sniffer),  or by any logs ?
>
> the best way to get info from teamd is using teamdctl util. For example:
>
> # teamdctl team0 state
> # teamdctl team0 state dump
>
> man teamdctl for more details.
>
> that should be enough to get the info you need. If you need anything
> else, please tell, the output can be easily extended.
>
> Jiri
> > Thanks,
> >
> > Andy
> > _______________________________________________
> > libteam mailing list
> > [e-mail]
> > https://lists.fedorahosted.org/mailman/listinfo/libteam

[Jiri Pirko]

Fri, May 30, 2014 at 05:02:00PM CEST, [e-mail] wrote:
> Hello,
>
> Thanks for your quick response.
>
> Still there is something strange when running "teamdctl team0 state",
> ,which bewilders me. I hope you can advice.
>
> I have a simple setup where I have a Linux server connected to Cisco
> 5000 ADswitch. I had configured LACP on two ports in the Linux server,
> and configured LAG on the Cisco 5000 switch.
> After running the teamd daemon with the LACP mode and running
> "teamdctl team0 state", I get data which shows that the LACP session
> is ok (so I understand, see below the full output).
>
> However, also after I remove the LAG from the switch and restart the
> daemon, I still get the same output from
> "teamdctl team0 state", which seems strange to me. I made sure that the
> LAG is not running on the switch, by "show port-channel database" (I
> get empty response).
>
> The LACP session should have not succeed in this case, so how come
> "teamdctl team0 state" is the same ? Or let's ask this: How does
> "teamdctl team0 state" should show that the LACP session was not
> established ?
>
> (or how "teamdctl team0 state dump" should be different ? )
>
> Following here is the full log of "teamdctl team0 state":
>
> teamdctl team0 state
> setup:
>  runner: lacp
> ports:
>  p5p1
>    link watches:
>      link summary: up
>      instance[link_watch_0]:
>        name: ethtool
>        link: up
>    runner:
>      aggregator ID: 0
>      selected: no
>      state: defaulted

When LACP link is established, aggregator is != 0, selected is
    yes for port of active aggregator and state is current.

So this output says that LACP link is not established
> p5p2
>    link watches:
>      link summary: up
>      instance[link_watch_0]:
>        name: ethtool
>        link: up
>    runner:
>      aggregator ID: 0
>      selected: no
>      state: defaulted
> runner:
>  active: yes
>  fast rate: yes
>
> Regards,
> Andy
>
> On Fri, May 30, 2014 at 4:01 PM, Jiri Pirko  wrote:
> > Fri, May 30, 2014 at 12:54:06PM CEST, [e-mail] wrote:
> > > Hello, libteam.
> > >
> > > When trying to establish a link aggregation with libteam, against an
> > > A/D switch (which is configured to support LACP), is there a way to
> > > know what is the status of the LACP session by some userspace tool
> > > (apart from a sniffer),  or by any logs ?
> >
> > the best way to get info from teamd is using teamdctl util. For example:
> >
> > # teamdctl team0 state
> > # teamdctl team0 state dump
> >
> > man teamdctl for more details.
> >
> > that should be enough to get the info you need. If you need anything
> > else, please tell, the output can be easily extended.
> >
> > Jiri
> > > Thanks,
> > >
> > > Andy
> > > _______________________________________________
> > > libteam mailing list
> > > [e-mail]
> > > https://lists.fedorahosted.org/mailman/listinfo/libteam