Release 2017.01 - RC2 about release-specs HOT 8 CLOSED

My functional tests for gcoap all passed on both native via TAP and samr21 via TUN (tunslip).

from release-specs.

08-Interop, Task 04 was successful with following results:

• payload 10B, interval 10ms 9% packet loss
• payload 100B, interval 100ms, 18% packet loss
• payload 1000B, interval 1000ms, 61% packet loss

NOTE: real wireless (15.4) connection between Raspberry Pi with Openlabs transceiver and SAMR21-Xpro, default channel settings (26, 0x23) distance approx. 2m.

Software:

• Raspberry Pi: radvd
• RIOT: GNRC networking

[edit]: btw. works in both directions RIOT <--> RasPi <--> Host

from release-specs.

08-Interop, Tasks 06 and 07 were successful, test with SAMR21-Xpro and RasPi as 6LBR

• UDP with 8B both directions
• UDP with 200B from host to RIOT only, because shell in GNRC networking has 128 char limit

from release-specs.

08-Interop, Task 05 successful, but somewhat unstable:

• ping between host and RIOT via RIOT 6LBR worked with approx. 8% packet loss
• 6LBR stops forwarding packets and does not respond to direct pings, though shell indicates no error on 6LBR
• had to try several times until pings got through and were forwarded correctly

from release-specs.

08-Interop, Task 05, second run more successful :)

• ping works in both directions, packet loss between 1 - 10%
• still: forwardings stopped (sometimes) for several seconds before self recovering

[edit] 6LBR was RIOT on samr21-xpro, RIOT node was pba-d-01-kw2x (was out of samr21-xpros)

from release-specs.

Thanks @kb2ma! @miri64 can you comment on @smlng's RIOT 6LBR results? Is this "known" issue? Do you remember results of your last release testing?

from release-specs.

Thanks @kb2ma! @miri64 can you comment on @smlng's RIOT 6LBR results? Is this "known" issue? Do you remember results of your last release testing?

Not exactly... At least the recovering part is new. But I don't know at the moment if I ever did these tests myself, either :-/. So I guess the safest thing is to check if it's reproducible on 2016.10.

from release-specs.

08-Interop, Task05, third run with 2 SAMR21-Xpros:

• works more smoothly in both directions
• RIOT -> RIOT 6LBR -> Host 0-1% packet loss
• Host -> RIOT 6LBR -> RIOT 5-10% packet loss

First observation, from host to RIOT exhibits (always) more packet loss. Second: using RasPi as 6LBR is more stable compared to RIOT-based 6LBR. Third: results very subjective to radio device and its antenna; especially when mixing different boards and radio devices.

However, so far results are reasonable. Btw. I checked against release 2016.10, similar results, so no (new) issues here.

from release-specs.