Move another 2-5m apart, and if poss have absorber inbetween - wall, window… get into range -65 to -95 if you can, higher or lower power levels should be ok, but helps remove another point of uncertainty if you can avoid any risk of RF front end distortion/saturation or bleed over…one of the main reasons why a node might not hear join accept even if on correct channels…
Which firmware version on the node?
When you registered the node id you use the “End device brand” feature or do it manually?
The RX2 got set where it ask you for frequency plan.
On a ready setup device you can check this in the console under Application → End Device → General Setting → Network Layer
And for RSSI as Jeff suggested, you will be better off getting it down.
Great thanks I added/selected them manually
both the gateway and the node are set to
Europe 863-870 MHz (SF12 for RX2)
Ive put the device in my car outside and moved it over the road to get the RSSI down to -88 to -90 and this does not seem to make any difference it is still in the same loop.
Firmware on the node (I think) is 1.7.0 I think this is the default, there is a slightly newer version (1.7.4) available but looks like you need a separate usb interface device to upgrade it.
and firmware on the gateway is 4.19.58
this is an older firmware too but if I put the newest firmware (4.2.5 - index - powered by h5ai v0.30.0 (https://larsjung.de/h5ai/)) on no matter how much I play around with the settings I cannot get the gateway to contact the things network anymore even if I update the details in the global config json file
So ignoring the line that says (SF9 for RX2) (Recommended)?!..
If you say that it joins when you are close to another gateway. And when at your own gateway when the RSSI is down to -88 to -90 it does not.
It could be the latency to the NS. Ping eu1 and see what it is, your pc needs to be on the same ethernet connection as your gateway.
eu1 - is in Ireland
I didn’t ignore the option RX2 SF12 was the default which was selected automatically so I left it as that.
Ive now set both to RX2 SF9 and this has not made any difference
Don’t worry about it you are on the optimum now…SF12 isn’t ‘default’ nothing shows to start - you click and its a drop down menu where that happens to be the first - each subgroup of frequency plans is listed alphabetically with IIRC “…MHZ (SF1xxxxxx…)” coming before “…MHz (SF9xxxxx…)” with the (Recommened) added to highlight prefered choice (2nd onlist) note further down where there are lots of options for sub bands even within subgroups of freq plans it gets even more explicit with again IIRC (as used on TTN). This is because experienced users can look to change and to tie in with private instances where other choices outside the community might be made… There are historic and technical reasons for why this choice was made when on the Community network. BTW where (approx.) are you located? Cant figure that stadium/sports ground in your picture 
sadly not at the Nou Camp which is where the picture is
I’m in the much less sunny DH4 near the Riverside Cricket Ground in Durham (Chester-le-Street) if you’re a sports fan
Cloudpingtest
Europe (Ireland) eu-west-1 120 ms
was-latency-test
Europe (Ireland) eu-west-1 32 ms
Cloudping
Europe (Ireland) eu-west-1 37 ms
Although the test results differ per site I’ve ran each one multiple times and the response times per site are pretty stable
The gateway seems to checkin with the network fine too, don’t know if that would be any different if I was experiencing latency issues
So your node is working now?
totally stumped now 
I left the node in my car overnight the node continued to to and check-in, the gateway would receive it, respond and this was not received by the node and the node would continue in the join request loop.
forgot all about it and drove to work at 7am and parked up at work by 7:30am
at 9:10am the note checked in, got a message back and send a data packet. even though there is not meant to be any coverage for some distance.
it then has checked in every hour at 10 past the hour all day. missed the checkin on my drive home but has completed the last three checkins successfully and I can see it on both the node and the gateway end so its definitely using my gateway and not another one near by.
nothing change on my gateway today or even since yesterday morning when I changed the RX settings
Never count on this, some gateways don’t have their location set, so you will never know.
These are normal uplinks.
Remember with a join there are three thing that need to happen (to simplify it). 1. uplink from node to LNS 2.LNS checking the credentials. 3. LNS sends downlink to confirm the join.
The downlink needs to occur in the window the node is on waiting for it, were its receiver is on. The new standard is 5 sec. The whole process of joining takes time, for the uplink, LNS and downlink to happen. That is why I said check the latency to eu1, not saying it were your problem. But have seen this causing issues with joining.
ill keep an eye on it for a few days and then start doing some range testing
In range of either the GW at Pity Me not far from the A1 or the intermittent GW near Bournmoor. Other possibilities are (if high enough in the hills -perhaps if driving around) over in Sunderland or the one on the East side of the Consett community…?
Once your node is picked up by a GW that can send a join accept that is received by the node ok all should then be good as the credentials are established and the node seen as ‘good’ by the LNS, node keeps transmitting messages that are picked up by any GW in range - and it looks like yours sees uplinks ok, and then forwards, so you would see traffic in your application even if you havent resolved the downlink reception issue associated with your GW… (I would expect!). If you fancy a bit of a drive anytime one of my GW’s (Wylam) over in the Tyne Valley is well visible from Crawcrook and much of Prudhoe - so you dont even have to cross the river 
you could try cycling your node there and see how well/easily it connects to the network…that would eliminate the node as a problem itself or highlight issues - the other option is get another COTS known good unit to test your GW and assist in debug - a TTN Node or a simple low cost T&H or door/window/contact sensor etc.
It sounds like the node firmware has a link check mechanism that gradually and slowly ratcheted up the SF over a few connection attempts until hitting one where the signal got through at a distance, which might explain why it finally linked once in a new place over those couple of hours…(roughly) where was work that day?
The standard for OTAA RX1 is fixed in the LoRaWAN regional specifications and has not changed.
That looks like there are 4 downlinks in a very narrow window which is suspect.
Can you please show you global_conf.json and local_conf.json?
wow @Jeff-UK didn’t realise you were so close.
My gateway is the one next to Bournmoor, should appear just at the entrance to our housing estate as I’m hosting it at home. I was ideally going to do some testing and get it set up with the eventual hope of upgrading the setup so that it could reach our allotment at Penshaw Monument so that i can do some remote monitoring.
Would be great to see your setup, I think i’m being fairly (over) ambitious in terms of range for non-commercial grade stuff. I do have another location I can use which is about half way between but is another house so would have to be a fairly small unobtrusive GW.
I have another couple of node units on order to do some more testing, initially looking at fairly basic stuff like temperature, humidity moisture type stuff but wanted to do it in a way that grew the network so others could use aswell rather than just doing a point to point thing for just my needs.
Good to know there are some more GW’s near by, the other ones i had been using to test is the one at FabLab in Sunderland and the one near my work (location unknown) near Silverlink in North Tyneside but obviously I cant see the backend of those ones
Actually not that close - Thames Valley based! with touch down locations in N.West and Tyne Valley. But I have some ~50 Gw’s deployed for community use around the country (and elsewhere) with another approx 120-150 through friends/collaborators & clients, all around the country.
Did some mapping & coverage checks around Netwon Aycliffe & Durham 5-7 years back when was looking to deploy GW’s and possibly set up community(ies) in N.East but got little traction. Did set up Tyne Valley Commmunity mind some years back and have 3 GW’s up there - 2 offline ATM - once since before Covid 
Get up a few times a year so may ping on Spring trip if you are around… due to put GW’s in Prudhoe & potentially Humshaugh near Hexham when I can sort backhaul arrangements…was looking on hills around A1 jn 61 & Durham services and down close by a curry house in Bowburn but site hosts wouldn’t commit to long term access! See there is now one intermittent over hills east side of A1 from jn 61 near Cassop (a TTIG monitoring a motorhome I believe?!)
have you done much with more domestic based gateways, guessing all of yours are full mast type deployments based on your issues with wayleaves and backhaul etc.
ideally i think i need to get about 1mile, but could probably get away with half a mile although there are houses and things in the way so wouldn’t have line of sight
@stroga, you’ve established that the device will join & send uplinks when in range of another gateway.
The gateway log you provided above doesn’t show any uplinks being heard.
So the most obvious test is to get the device about 5m + brick wall away from the gateway, turn it on and then go and look at the gateway log on the Pi to see if it is hearing the device.
And in the spirit of “is it plugged in”, is there an antenna on the gateway?
Will DM you rather than take up this thread with more off topic discussion