Node-red crashes at startup with 2.1.2 addMembership EADDRINUSE

[0rsa]

** Describe issue **
Just updating from node-red-contrib-miio-roborock 2.0.7 to 2.1.2, node-red crashes at startup, even with a fresh system restart

** Screens / Logs **

27 Mar 04:21:12 - Error: addMembership EADDRINUSE
at Socket.addMembership (dgram.js:828:11)
at /root/.node-red/node_modules/node-mihome/lib/protocol-aqara.js:48:26
at Array.forEach ()
at /root/.node-red/node_modules/node-mihome/lib/protocol-aqara.js:46:26
at Array.forEach ()
at Socket. (/root/.node-red/node_modules/node-mihome/lib/protocol-aqara.js:44:38)
at Socket.emit (events.js:327:22)
at Socket.EventEmitter.emit (domain.js:467:12)
at startListening (dgram.js:149:10)
at dgram.js:344:7
at processTicksAndRejections (internal/process/task_queues.js:81:21)

** Hardware / Software **
Debian Buster
Node-red 1.2.9

Downgrading to 2.0.7, node-red starts fine.

[andreypopov]

nodejs version?

[0rsa]

@andreypopov 14.16.0

[0rsa]

I tried with 15.12.0, same behaviour:

27 Mar 19:06:56 - [info] Node-RED version: v1.2.9
27 Mar 19:06:56 - [info] Node.js version: v15.12.0
27 Mar 19:06:56 - [info] Linux 4.19.0-14-amd64 x64 LE
27 Mar 19:06:57 - [info] Loading palette nodes
SmartthingsConfigNode
27 Mar 19:06:58 - [red] Uncaught Exception:
27 Mar 19:06:58 - Error: addMembership EADDRINUSE
at Socket.addMembership (node:dgram:849:11)
at /root/.node-red/node_modules/node-mihome/lib/protocol-aqara.js:48:26
at Array.forEach ()
at /root/.node-red/node_modules/node-mihome/lib/protocol-aqara.js:46:26
at Array.forEach ()
at Socket. (/root/.node-red/node_modules/node-mihome/lib/protocol-aqara.js:44:38)
at Socket.emit (node:events:381:22)
at Socket.emit (node:domain:470:12)
at startListening (node:dgram:168:10)
at node:dgram:363:7
at processTicksAndRejections (node:internal/process/task_queues:82:21)

[andreypopov]

looks like a conflict, do you have nodes connected with xiaomi ? give me the list

[0rsa]

"dependencies": {
"miio": "github:kingkong123/miio",
"moment": "^2.29.1",
"node-mihome": "0.0.27",
"node-red-contrib-contextbrowser": "0.0.4",
"node-red-contrib-deconz": "~1.3.2",
"node-red-contrib-eesmart-d2l": "~0.1.0",
"node-red-contrib-google-smarthome": "~0.0.63",
"node-red-contrib-miio-roborock": "~2.0.8",
"node-red-contrib-multipart-stream-decoder": "0.0.4",
"node-red-contrib-nanoleaf-aurora": "~0.2.2",
"node-red-contrib-netatmo": "~0.2.0",
"node-red-contrib-nexmo": "~3.3.1",
"node-red-contrib-rfxcom": "~2.11.3",
"node-red-contrib-smartthings": "0.0.20",
"node-red-contrib-stackhero-mysql": "~1.0.5",
"node-red-contrib-sunevents": "~3.0.3",
"node-red-contrib-wled": "0.0.4",
"node-red-contrib-yeelight": "~1.0.7",
"node-red-contrib-zwave-js": "~3.1.2",
"node-red-gardena": "file:../../home/flo/node-red-modules/node-red-gardena",
"node-red-node-ping": "~0.3.0",
"node-red-node-rbe": "~0.5.0",
"node-red-node-suncalc": "~1.0.1",
"node-red-node-tail": "~0.3.0",
"openzwave-shared": "^1.7.2",
"passport-google-oauth": "^2.0.0",
"sodium": "^3.0.2",
"suncalc": "^1.8.0"
}

[0rsa]

I removed miio and node-mihome which are old dependencies. Error persists.

[0rsa]

Ok so I commented api.js l.6 and it starts :
mihome.aqaraProtocol.init();

Why you need to init zigbee connection for roborocks?
Also, even with 2.2.2, I have exceptions which cause node red to crash completely.

28 Mar 07:00:59 - [warn] [miio-roborock-server:S7] Miio Roborock Error: Could not connect to device, handshake timeout
This error originated either by throwing inside of an async function without a catch block, or by rejecting a promise which was not handled with .catch(). The promise rejected with the reason:
Error: Could not connect to device, handshake timeout
at Timeout._onTimeout (/root/.node-red/node_modules/node-red-contrib-miio-roborock/node_modules/miio/lib/network.js:427:17)
at listOnTimeout (node:internal/timers:557:17)
at processTimers (node:internal/timers:500:7)

It started with 2.0.7 a few days ago, my node red crashs a lot of times per day due to node-red-contrib-miio-roborock.

---

Edit: as it's a warning only, maybe this last crash (first using 2.2.2) is due to something else. Server is up and running, I will see in next hours if it keeps crashing due to node-red-contrib-miio-roborock or not.

I have an idea about EADDRINUSE, I use two roborocks, maybe can it be related to 2 instances of aqara initiated?

[becmar39]

Maybe the same as in #19

[mobamoba]

I'm having the same issue with 2.1.2 and I can see that the address conflict is with port 9898 which, in my case, is the port used by the Xiaomi Gateway. Perhaps knowing that will help lead to a solution.

[andreypopov]

Please, test this version

[email protected]

[dprokscha]

Well, yes. I am the guy from #38 and I have the same issue. Downgrading to 2.0.7 solves the issue for me too. I am encountering this error on a fresh clean install of Node-RED. My environment:

• Raspberry Pi 4, 4 GB RAM, static IP (wired), running Ubuntu Server 21.10 (64-bit)
• Node.js 14.18.1
• Node-RED 2.1.3

As mentioned: It is a fresh clean install - no other modules are installed (except of the Pi specific ones which were installed by default).

I did some investigations and found out, that this error does not occur if I am using Ubuntu Server 20.04 LTS (64-bit). So maybe it is related to the underlaying OS. I also doublechecked if any ports (especially 9898) are already in use - but they aren't. The big questions are:

• What changed from Ubuntu 20 to 21?
• What changed from node-red-contrib-miio-roborock 2.0.7 to 2.3.0 (or in its libraries)?
• What of these changes causes the conflict?

I don't know where to start but maybe this hint with the underlaying OS is enough to help @andreypopov debugging this issue. Please let me know if you need any further information 😜

[mobamoba]

I haven't upgraded my OS (18.04 here) and have the same problem so thinking an OS upgrade isn't the issue.

[andreypopov]

I have ubuntu too.(
I tried to create 5 instances of configuration, so I had 5 connects to roborock and everything worked well.
Do you have other plugins which are connected with mihome, aqara ?
Do you have telegram? can u write to me and we will debug this issue together? @andrey_p3

[dprokscha]

Hm, now I was able to update to 2.3.1 😂

The only thing what changed: I noticed multiple IPs for my network interface "eth0" - one static as mentioned above and another dynamic one. The last one exists because of a misconfiguration of netplan. I deleted the property "dhcp4" from the configuration to get my static IP. But the right way is to set it to "false". Otherwise you'll get the unwanted dynamic IP, because the default value from "dhcp4" seems to be "true".

I am supposing that Node-RED/miio is not handling multiple IPs the right way. It tries to bind the server port 9898 multiple times. Maybe this leads to the error "EADDRINUSE".

@mobamoba: Can you please share your output for "ip address show" to doublecheck my assumption?

[mobamoba]

Sure:

1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: wlp2s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
link/ether 10:02:b5:31:cd:92 brd ff:ff:ff:ff:ff:ff
inet 192.168.1.46/24 brd 192.168.1.255 scope global noprefixroute wlp2s0
valid_lft forever preferred_lft forever
inet6 fe80::1202:b5ff:fe31:cd92/64 scope link
valid_lft forever preferred_lft forever

[dprokscha]

This is really interesting. Your network configuration seems to be OK. The only difference to mine is that you are using WIFI only.

The root cause in my case is definitely the configuration of netplan and the unwanted dynamic IP via "dhcp4". If I toggle "dhcp4" to "true" and I have both, a static and a dynamic IP, I can reproduce the error. If "dhcp4" is "false" (static IP only), everything is fine and is working as expected.

In your case @mobamoba something different is consuming the port 9898. Maybe it is another module, another instance of Node-RED, another process or Node-Red/miio has trouble with your network configuration in some way.

But I have no clue to be honest - sounds like you need a debugging session with Andrey...

[mobamoba]

I use a different Xiaomi node to control my Xiaomi Gateway and all my Aqara sensors. The problem, for me, is that the later versions of this app included something that forced the use of 9898 which is impossible in my system and is actually unnecessary as I'm using 2.0.7 - which doesn't have the 9898 conflict - and it controls my Roborock perfectly.

I think the solution lies in removing whatever got added that is now using 9898 and remove that. What exactly is in this node that's using 9898 now and that wasn't using it in 2.0.7?

[B0rax]

Ok so I commented api.js l.6 and it starts : mihome.aqaraProtocol.init();

Just fyi. This Issue here is still in the newest version. I did uncomment the line that you mentioned and it fixed the issue for me.

@andreypopov what does this line do? In my testing (a few months now) it seems to work just fine without this line.

[mobamoba]

I'm pretty sure the issue is here in node mihome:

https://github.com/maxinminax/node-mihome/blob/master/lib/protocol-aqara.js
const MULTICAST_ADDRESS = '224.0.0.50';
const MULTICAST_PORT = 4321;
const SERVER_PORT = 9898;

If port 9898 is in use, as it is in my case, this node fails and prevents node-red from loading. I don't know about the multicast stuff.

[DeadEnded]

I just had this happen today without any changes in NR or pallet, etc. after a restart of the container.
I did some digging and found MY cause - which might shed some light on this and help fix it.

Firstly, my error looked like this (log from portainer - might not be as detailed, but it was enough):

10 May 10:06:19 - [red] Uncaught Exception:
10 May 10:06:19 - [error] Error: addMembership EADDRINUSE
    at Socket.addMembership (dgram.js:850:11)
    at /data/node_modules/node-mihome/lib/protocol-aqara.js:48:26
    at Array.forEach (<anonymous>)
    at /data/node_modules/node-mihome/lib/protocol-aqara.js:46:26
    at Array.forEach (<anonymous>)
    at Socket.<anonymous> (/data/node_modules/node-mihome/lib/protocol-aqara.js:44:38)
    at Socket.emit (events.js:412:35)
    at startListening (dgram.js:172:10)
    at dgram.js:364:7
    at processTicksAndRejections (internal/process/task_queues.js:83:21)
npm ERR! code ELIFECYCLE
npm ERR! errno 1
npm ERR! [email protected] start: `node $NODE_OPTIONS node_modules/node-red/red.js $FLOWS "--userDir" "/data"`
npm ERR! Exit status 1

As @mobamoba pointed out, the protocol-auara.js appears to be part of the issue.
Looking at this class, there are indeed some static ports. For me though, this might not have been the issue.
My error points to Socket.addMembership called from protocol-aquara.js line 48.
I can see that the init that was commented out to fix this, calls the createSocket.
The createSocket is right below it, and contains the code where the error is happening.

Now, I see that in this line it is specifically looking for Multicast_Address - nothing to do with the ports.
I scratched my head for a bit... then I realized last week I was messing around with adding a second IP to my system.
So I checked out netplan and sure enough I had not removed my testing.
I had defined a second IP for my ethernet port.
I removed this, restarted Node-Red, and it's fixed!

So - I know that was a lot of information - but I hope my digging into this might help shed light on what is really causing the issue.
For me, it was having two IP addresses defined.
I assume the code opened the port on the first address, and then when it tried to do it again on the second, it threw the error.
How to gracefully handle this - that is outside of my knowledge currently.
It does appear to be a bug on the protocol-auara.js though.

I'll going to submit this as an issue to that repo - maybe it will help them fix this!

Cheers!
DeadEnd