direct messages NextHopRouting between nodes with 2.6.11 and >= 2.7.11

[shalberd]

Hi @GUVWAF @mikecarper

I and a few others here in Switzerland are trying out the alpha 2.7 firmwares, starting from 2.7.11 and now 2.7.12.

We have noticed, our mountain ROUTER_LATE nodes running on 2.6.11 and most other client nodes, too, that direct messages do not arrive at the end recipient anymore. Nor do ACKs come back.

I have noticed your changes in the code and the many great improvements. When 2.7 becomes stable / beta, it will be a great improvement in terms of direct messages reliability.

For now, can you just answer whether indeed firmware 2.6.11 and firmware starting from 2.7.11 cannot send each other end to end direct messages anymore?

[GUVWAF]

For now, can you just answer whether indeed firmware 2.6.11 and firmware starting from 2.7.11 cannot send each other end to end direct messages anymore?

No, that shouldn't be the case; all the changes are backwards compatible.
If you experience issues then we'd need to figure out what is going wrong, but for that we'd need more information like serial logs.

[shalberd]

@GUVWAF give me some time, probably 2-3 days. I will attach correct serial logs with debug info etc.
This evening, I seem to have a larger issue with firmware 2.7.12. Suddenly, no messages get out and first node ACKs anymore, even on Flood Router. That has nothing to do with 2.7.12 though, as MEDIUM_FAST worked up to 20 minutes ago.

[GUVWAF]

The debug logs show they do not have set a next-hop yet: "Setting next hop for packet with dest d8aa5141 to 0" (0 means no next-hop). But it also shows the node does not hear anyone rebroadcasting the packet.

Do channel messages work normally?

[shalberd]

Yeah, sorry, no, since about 25 minutes, channel messages do not work normally, either. It wasn't an issue earlier today on channel messages, will get back to you here soon.

[GUVWAF]

Ah, my previous comment was about your initial debug logs. I'll wait for the new ones.

[shalberd]

OK, so I can send out channel messages again:

Now, when I send direct message out, I get the following, cloud with strikethrough window:

Debug log:
serial log 2_7_12.txt

[GUVWAF]

You seem to be sending fine, but you get a NAK back because of PKI failure:
"INFO | 18:53:13 1000 [Router] PKI decrypt failure, send a NodeInfo".

If you factory reset your node, the other side has to delete your node from their NodeDB, and then you can try again.

[shalberd]

You seem to be sending fine, but you get a NAK back because of PKI failure:
"INFO | 18:53:13 1000 [Router] PKI decrypt failure, send a NodeInfo".
If you factory reset your node, the other side has to delete your node from their NodeDB, and then you can try again.

Yes, that could be it, definitely. I DID do a factory reset when I updated from 2.6.11 to 2.7.11.
To tell the other side the new keys, is it also possible for me to do a "Exchange User Info" and that way renew my public key info on their side, or does the other side always have to delete the node and wait for it to re-appear?

[GUVWAF]

Yes, the other side always has to delete the node, as we don't accept new keys (since it could be someone is trying to impersonate that node). We're working on improving this though: #8211 (comment)

[shalberd]

Awesome, I will keep you apprised over the course of the week here, probably from Tuesday or Wednesday on. Good to know though that it is most likely not a 2.7.12 vs. 2.6.11 problem.
I also communicated this info, having to delete nodes from node DB if direct messaging does not work e.g. after factory reset, here in our MEDIUM_FAST main channel in Switzerland.