LAG LACP timeout tunable through IOCTL

[lakshmi.n_msystechnologies.com]

Summary

In FreeBSD-10 and in Current, by default LACP supports only long timeout.
FreeBSD does not provide the way to configure LACP timeout period.
We made code changes for LACP Fast-timeout (Using IOCTL, both GET / SET) on FreeBSD-11.
And we were able to successfully test the operation using IOCtl calls from userland.

Please let us know your comments to take this forward.

Test Plan

Introduction:

Test plan for LACP timeout tunable through IOCTL

Prepared by:

Lakshmi Narasimhan Sundararajan

Testing Strategy

Unit testing:

1. Form a LACP between 2 machine say (A) and (B). (A) is development machine and (B) to support functionality of LACP
2. Ensure by default, machine (B) sends LACP PDU in slow timeout.
3. Using ifconfig utility, tune machine (A) LACP timeout value to fast timeout.
4. After tuning ensure that machine (B) honour fast timeout for machine (A).

Dependencies:

Changes require on userland ifconfig and if_lagg (LAGG) driver

Documentation:

Updated the ifconfig.8 man page

[wblock]

Please see https://wiki.freebsd.org/CodeReview#Create_a_Revision_via_Web_Interface about including more context. Thanks!

sbin/ifconfig/ifconfig.8

2400

Needs an ending period. "on the interface" might be unnecessary, it's hard to tell.

2402

Needs an ending period. "on the interface" might be unnecessary, it's hard to tell.

[lakshmi.n_msystechnologies.com]

Thanks warren,

Addressed your review comments on:

1. Ending period changes on man page
2. Uploaded svn diff with full context

[rpokala-panasas.com]

Looks good to me.

[lakshmi.n_msystechnologies.com]

Warren, I have addressed your comments. Please let me know how I can speed up the process to get this integrated to mainline.

[hiren]

Looks okay to me.

[smh]

The option used here isn't ideal IMO as it's meaning, which is to enable / disable "fast" timeout, isn't obvious and could easily be miss-interpreted without reading the man page.

Using lacp_fast_timeout -lacp_fast_timeout, while a little longer, would make obvious from the flag name.

[gnn]

Agreed on the name change, but then it's fine to commit IMHO.

[lakshmi.n_msystechnologies.com]

Thanks Hiren / Steven / George for your review comments.

Addressed the review comments on name change.

[smh]

Thanks for making that change :)

[rpokala-panasas.com]

Thanks LN.

[smh]

All looks good to me now, time to get this in?

[rpokala]

In D3300#91230, @smh wrote:

All looks good to me now, time to get this in?

Actually, this went in already, in r286700, but it wasn't tagged w/ this review. Marking this closed now.

Hmm, except when I try to close it, I get: "You can not close this revision because it has not been accepted. You can only close accepted revisions."

Ah, it looks like it's still marked red for "manpages"...

[bjk]

The man pages are fine.