[chrony] unix socket connection is broken

[Frankkkkk]

Relevant telegraf.conf

[[inputs.chrony]]
    server = "unix:///var/run/chrony/chronyd.sock"
    metrics = ["activity", "tracking", "sources", "sourcestats"]

Logs from Telegraf

[pid 612402] write(9, "\6\1\0\0\0,\0\0\0\0\0\1\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 416 <unfinished ...>
[pid 612402] <... write resumed>)       = 416
[pid 612402] read(9,  <unfinished ...>
[pid 612402] <... read resumed>0xc003368400, 1024) = -1 EAGAIN (Resource temporarily unavailable)

System info

1.13 - 78cbf533d2

Docker

No response

Steps to reproduce

1. Use the chrony plugin
2. Set the server to be unix:///var/run/chrony/chronyd.sock
3. Launch telegraf as root (or at least as a user that con use the unix socket)
4. See it hang

Expected behavior

The chrony plugin should work

Actual behavior

The plugin hangs when using the unix socket when:

• telegraf is launched as root
• or the connection is manually specified to use the socket

The reasons are two fold:

• Chrony uses datagram (unixgram) unix socket and not stream (unix) ones.
• The socket should be duplicated in order not to hang other connections to it

Here is a working example by facebook

Additional info

I guess this is hard to reproduce, because the unix socket is rarely used as telegraf cannot talk to it (thus it instead connects via UDP to the chrony deamon).

The unix socket is however needed to access the serverstats metrics. Most of the other ones can be reached using the udp socket.

[powersj]

Thanks for the various issues and PRs

next steps: review PR