So just increasing the conn itvl hasn't solved it, which makes me suspect
it isn't simply a timing thing, but rather a peer not responding to the
request (or responding in an unexpected way). I tried up to 180ms, which
should give it a full second to receive the first data frame.

014762 [ts=115328104ssb, mod=4 level=1] GAP procedure initiated: connect;
peer_addr_type=1 peer_addr=42:62:84:90:38:f7 scan_itvl=16 scan_window=16
itvl_min=24 itvl_max=144 latency=1 supervision_timeout=512 min_ce_len=16
max_ce_len=768
014792 [ts=115562464ssb, mod=64 level=1] peer: connected; conn_handle=29
status=0 addr=42:62:84:90:38:f7
014794 [ts=115578088ssb, mod=4 level=1] GATT procedure initiated: exchange
mtu
014907 *** ble_ll_conn.c:2160 *** cputime=116849198 anchor_point=117028850
last_rxd_pdu_cputime=115946959 tmo=1080000
014910 [ts=116484344ssb, mod=64 level=1] peer: updated; conn_handle=29
status=520 itvl=144 latency=1 tmo=512
014912 [ts=116499968ssb, mod=64 level=2] peer: failed to exchange mtu;
conn_handle=29 status=7
014915 [ts=116523404ssb, mod=64 level=1] peer: disconnected; conn_handle=29
reason=520

I will try the 1.2 branch and report.


On Tue, Sep 5, 2017 at 12:43 PM, will sanfilippo <[email protected]> wrote:

> Glad other folks were answering you Simon; I was away for a bit.
>
> I do not have anything else to add. The only thing is that if you do not
> have a sniffer one (possible) way to determine what is going on is to look
> at statistics. You need to look at the stats directly before and after the
> issue which could be a problem. What that will tell you is if the
> peripheral heard the connect request from the master and to determine which
> side is not hearing the data frames. If you want to go that route please
> let me know and I can tell which stats to look for. This might be difficult
> for you to set up if it is that infrequent.
>
> I would also try out the new code base as well if that is easily doable
> for you. Many things were fixed/changed so using that codebase going
> forward would certainly be best (imo)
>
>
> > On Sep 5, 2017, at 11:57 AM, Łukasz Rymanowski <
> [email protected]> wrote:
> >
> > Hi,
> >
> > On Sep 5, 2017 8:15 PM, "Simon Ratner" <[email protected]> wrote:
> >
> > On Tue, Sep 5, 2017 at 11:01 AM, Łukasz Rymanowski <
> > [email protected]> wrote:
> >
> >>
> >> Note that this is how BLE works. Master sends LE Create Connection on
> >> Advertising event and assumes connection is created. In this point of
> time
> >> host gets Connection Complete Event according to BT spec. However, for
> the
> >> reasons Will described, it might happen that peer will not answer on any
> > of
> >> the first 6 connection events. In this case connection is considered as
> >> dropped and  Disconnect Complete Event will come with "Connection failed
> > to
> >> be established" goes to host
> >
> >
> > Got it, thanks Łukasz! I am sure that the timeout is happening in exactly
> > this place.Since it seems that in all cases the times are very close, i
> > think i can increase connection interval to make this significantly less
> > frequent.
> >
> >
> > Please let us know the result.
> > BTW Air logs would help to confirm what is happening.
> >
> > \Łukasz
>
>

Reply via email to