phpfpm plugin with unix socket fails

[sofixa]

Hello, everybody.

Awesome work on telegraf, it's really a pleasure to use it, and i've had no issues with it whatsoever until recently.

I setup a webserver with nginx and php5-fpm, with communcation between them passing via the unix socket at /var/run/php5-fpm.sock.

Whenever i run telegraf, with -test and -debug, i get a rather long stacktrace, but with nothing that actually helps me debug:

* Plugin: phpfpm, Collection 1
panic: runtime error: index out of range

goroutine 22 [running]:
github.com/influxdb/telegraf/plugins/phpfpm.(*phpfpm).gatherServer(0xc8201c23c0, 0xc8201c22e1, 0x16, 0x7f9431fa77a0, 0xc8201d2380, 0x0, 0x0)
        /home/telebuild/go/src/github.com/influxdb/telegraf/plugins/phpfpm/phpfpm.go:138 +0x731
github.com/influxdb/telegraf/plugins/phpfpm.(*phpfpm).Gather.func1(0xc8201d13b0, 0xc8201d13c0, 0xc8201c23c0, 0x7f9431fa77a0, 0xc8201d2380, 0xc8201c22e1, 0x16)
        /home/telebuild/go/src/github.com/influxdb/telegraf/plugins/phpfpm/phpfpm.go:83 +0x7b
created by github.com/influxdb/telegraf/plugins/phpfpm.(*phpfpm).Gather
        /home/telebuild/go/src/github.com/influxdb/telegraf/plugins/phpfpm/phpfpm.go:84 +0x1d1

goroutine 1 [runnable]:
sync.runtime_Semacquire(0xc8201d13bc)
        /usr/local/go/src/runtime/sema.go:43 +0x26
sync.(*WaitGroup).Wait(0xc8201d13b0)
        /usr/local/go/src/sync/waitgroup.go:126 +0xb4
github.com/influxdb/telegraf/plugins/phpfpm.(*phpfpm).Gather(0xc8201c23c0, 0x7f9431fa77a0, 0xc8201d2380, 0x0, 0x0)
        /home/telebuild/go/src/github.com/influxdb/telegraf/plugins/phpfpm/phpfpm.go:87 +0x1fe
github.com/influxdb/telegraf.(*Agent).Test(0xc820028258, 0x0, 0x0)
        /home/telebuild/go/src/github.com/influxdb/telegraf/agent.go:197 +0x504
main.main()
        /home/telebuild/go/src/github.com/influxdb/telegraf/cmd/telegraf/telegraf.go:115 +0xe3c

goroutine 17 [syscall, locked to thread]:
runtime.goexit()
        /usr/local/go/src/runtime/asm_amd64.s:1696 +0x1

goroutine 5 [syscall]:
os/signal.loop()
        /usr/local/go/src/os/signal/signal_unix.go:22 +0x18
created by os/signal.init.1
        /usr/local/go/src/os/signal/signal_unix.go:28 +0x37

goroutine 11 [runnable]:
github.com/influxdb/telegraf.(*Agent).Test.func1(0xc820016ea0, 0xc820016e40)
        /home/telebuild/go/src/github.com/influxdb/telegraf/agent.go:178 +0xc0
created by github.com/influxdb/telegraf.(*Agent).Test
        /home/telebuild/go/src/github.com/influxdb/telegraf/agent.go:185 +0xd6

goroutine 20 [IO wait]:
net.runtime_pollWait(0x7f9431fa87d0, 0x72, 0xc82000e1e0)
        /usr/local/go/src/runtime/netpoll.go:157 +0x60
net.(*pollDesc).Wait(0xc820173fe0, 0x72, 0x0, 0x0)
        /usr/local/go/src/net/fd_poll_runtime.go:73 +0x3a
net.(*pollDesc).WaitRead(0xc820173fe0, 0x0, 0x0)
        /usr/local/go/src/net/fd_poll_runtime.go:78 +0x36
net.(*netFD).Read(0xc820173f80, 0xc8201cb000, 0x1000, 0x1000, 0x0, 0x7f9431fa2050, 0xc82000e1e0)
        /usr/local/go/src/net/fd_unix.go:232 +0x23a
net.(*conn).Read(0xc8200282c0, 0xc8201cb000, 0x1000, 0x1000, 0x0, 0x0, 0x0)
        /usr/local/go/src/net/net.go:172 +0xe4
net/http.noteEOFReader.Read(0x7f9431fa8c08, 0xc8200282c0, 0xc8201434f8, 0xc8201cb000, 0x1000, 0x1000, 0x0, 0x0, 0x0)
        /usr/local/go/src/net/http/transport.go:1370 +0x67
net/http.(*noteEOFReader).Read(0xc8201c39e0, 0xc8201cb000, 0x1000, 0x1000, 0x0, 0x0, 0x0)
        <autogenerated>:126 +0xd0
bufio.(*Reader).fill(0xc820017380)
        /usr/local/go/src/bufio/bufio.go:97 +0x1e9
bufio.(*Reader).Peek(0xc820017380, 0x1, 0x0, 0x0, 0x0, 0x0, 0x0)
        /usr/local/go/src/bufio/bufio.go:132 +0xcc
net/http.(*persistConn).readLoop(0xc8201434a0)
        /usr/local/go/src/net/http/transport.go:876 +0xf7
created by net/http.(*Transport).dialConn
        /usr/local/go/src/net/http/transport.go:685 +0xc78

goroutine 21 [select]:
net/http.(*persistConn).writeLoop(0xc8201434a0)
        /usr/local/go/src/net/http/transport.go:1009 +0x40c
created by net/http.(*Transport).dialConn
        /usr/local/go/src/net/http/transport.go:686 +0xc9d

My telegraf config is regular, as per its doc:

[[plugins.phpfpm]]
  urls = ["/var/run/php5-fpm.sock"]

The sock is present, and is with 666 rights, so telegraf should be able to access it.
The script that i found here works perfectly, so i suppose i am missing something between Telegraf and my sock rather than a php-fpm config issue.

What can I do so that Telegraf successfully connects to my sock? Thanks in advance to anybody who responds to my plead for help :)

PS: I tried enabling the pm.status variable in my php-fpm pool config, to /fpmstatus, but i am running a WordPress blog with a lot of redirections(i18n, cache, cdn) and my requests get redirected to the website and fail with a 404, so using the web to gather stats by Telegraf isn't an option, that's why i am trying to do it via the sock.

[sofixa]

In fact, i just discovered my problem - the path to the sock should be prefixed by localhost:(so localhost:/var/run/php5-fpm.sock instead of just /var/run/php5-fpm.sock). That isn't very clear in the documentation of the plugin, so i propose that you edit it to mention that little detail.

[sparrc]

thanks for the detailed report 👍

[v9n]

Will submit PRs for this.

[sparrc]

thanks @kureikain

[stvnwrgs]

For me it is still not working.

My config:

[[inputs.phpfpm]]
    urls = ["127.0.0.1:/var/run/php5-fpm.sock"]

2016/03/30 17:01:45 Error in input [phpfpm]: Socket doesn't exist  '127.0.0.1': stat 127.0.0.1: no such file or directory

[nsteinmetz]

@stvnwrgs

I had the same issue ; it was a little bit improved when I add a real bash to telegraf user (instead of /bin/false ; otherwise it failed to read the socket).

From #502, I added:

[[inputs.phpfpm]]
    urls = ["/var/run/php5-fpm.sock:status"]

If you look at the code, it's

    urls = ["`path_to_socket<:optional status function>"]

But I still have:

Error in input [phpfpm]: Socket doesn't exist  'localhost': stat localhost: no such file or director

Whereas I defined a socket. So arguments are not well parsed ?

What should be perms on socket? Seems 0666 is not enough?