TypeError: Cannot read property 'close' of null

[dwelle]

Happens in cluster mode upon server restart.

pm2 version 0.15.9, Node v4.2.1, Windows 7.

test-1 TypeError: Cannot read property 'close' of null
test-1     at Worker.disconnect (cluster.js:664:19)
test-1     at Worker.onmessage (cluster.js:529:16)
test-1     at process.<anonymous> (cluster.js:710:8)
test-1     at emitTwo (events.js:92:20)
test-1     at process.emit (events.js:172:7)
test-1     at handleMessage (internal/child_process.js:686:10)
test-1     at Pipe.channel.onread (internal/child_process.js:440:11)

config:

"name": "test",
"script": "./server.js",
"watch": "./",
"autorestart": false,
"exec_mode": "cluster_mode",
"instances": 2

[dwelle]

Suddenly can't reproduce. Will post here if it returns.

[indreek]

I have the same problem on ubuntu

config {
"apps" : [{
// Application #1
"name" : "XXXX",
"script" : "server.js",
"merge_logs" : true,
"cwd" : "/opt/api/XXXX",
"instances" : 0,
"exec_mode" : "cluster_mode",
"error_file" : "/opt/api/logs/pm2/error.log",
"out_file" : "/opt/api/logs/pm2/out.log",
"pid_file" : "/opt/api/pid/XXXX.pid"
}]
}

when i do pm2 stop all i get error

XXXX-0 (err): TypeError: Cannot read property 'close' of undefined
XXXX-0 (err): at Worker.disconnect (cluster.js:664:19)
XXXX-0 (err): at Worker.onmessage (cluster.js:529:16)
XXXX-0 (err): at process. (cluster.js:710:8)
XXXX-0 (err): at emitTwo (events.js:92:20)
XXXX-0 (err): at process.emit (events.js:172:7)
XXXX-0 (err): at handleMessage (internal/child_process.js:685:10)
XXXX-0 (err): at Pipe.channel.onread (internal/child_process.js:440:11)
XXXX-0 (err): TypeError: Cannot read property 'close' of undefined
XXXX-0 (err): at Worker.disconnect (cluster.js:664:19)
XXXX-0 (err): at Worker.onmessage (cluster.js:529:16)
XXXX-0 (err): at process. (cluster.js:710:8)
XXXX-0 (err): at emitTwo (events.js:92:20)
XXXX-0 (err): at process.emit (events.js:172:7)
XXXX-0 (err): at handleMessage (internal/child_process.js:685:10)
XXXX-0 (err): at Pipe.channel.onread (internal/child_process.js:440:11)

[dwelle]

Happened to me once again when I mistakenly ran two Hapi server connections on the same port. Pm2 probably assumes them identical when restarting, and the 2nd throws when trying to close an already closed connection.

[soyuka]

Both stack traces are linking to this line.

Does pm2 fail after having this error or only the script? If you had a way to reproduce it'd be lovely, thanks!

[dwelle]

The server is killed:

PM2 App name:app id:1 disconnected
PM2 8924 pid can not be killed { [Error: kill ESRCH] code: 'ESRCH', errno: 'ESRCH', syscall: 'kill' }
PM2 Starting execution sequence in -cluster mode- for app name:app id:1
PM2 App name:app id:1 exited with code 100
PM2 App name:app id:1 online
app-0 Server started on http://app:3000
app-1 Server started on http://app:3000

[Unitech]

closing in favor of #2061