fix(host_metrics source): fix tcp netlink bug

[mushrowan]

Summary

fix tcp metrics netlink parsing error (#22487)

fixes the "Could not parse netlink response: Decode error occurred: invalid
netlink buffer" error that occurs when collecting tcp metrics. this was caused
by socket.recv() with a fixed 4096 byte buffer but didn't track how many bytes
were actually received.

changed to use socket.recv_from_full() which returns Vec<u8> containing only
the actual received bytes.

Vector configuration

[sources.host_metrics]
type = "host_metrics"
collectors = ["network"] # for tcp

How did you test this PR?

# run tcp metrics tests (including ignored integration test)
cargo test -p vector --no-default-features --features sources-host_metrics tcp -- --include-ignored

all tests pass. Have also made a docker image @ tocalabs0509/vector:dev, which
works including the new bits. Note that it's a nix container I made for
testing just these changes, so other stuff e.g. journald doesn't work on it, but
should work in the final image.

Change Type

• Bug fix
• New feature
• Non-functional (chore, refactoring, docs)
• Performance

Is this a breaking change?

• Yes
• No

Does this PR include user facing changes?

• Yes. Please add a changelog fragment based on our
  guidelines.
• No. A maintainer will apply the no-changelog label to this PR.

References

• Closes: host_metrics - linux - Invalid netlink buffer length #22487

Files Changed

• src/sources/host_metrics/tcp.rs - fixed netlink receive to use
  recv_from_full()

Extra note

I was building this on my system using Nix, for which I wrote a flake.nix. I
assumed that people wouldn't want a flake.nix and flake.lock clogging up the
root of the repo so I haven't included it, but if this is wanted I'm happy to
add to the PR.
https://github.com/mushrowan/vector-flake in case you want to look.

[github-actions]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[mushrowan]

I have read the CLA Document and I hereby sign the CLA

[mushrowan]

hmm i guess i have to split this into 2 prs - will do in a bit

[mushrowan]

looks like grafana is interpreting the new inode metrics as counters for some reason, will double check the format in a bit

[mushrowan]

Have now split the inodes bit onto another branch which i will PR separately.

[mushrowan]

looks like grafana is interpreting the new inode metrics as counters for some reason, will double check the format in a bit

I think this is just grafana getting confused when you only have one host with a metric and the metric keeps going up. the format of the JSON is fine by the looks of it - i suspect the warning will disappear once you have more than one host (warning also doesn't appear on the inode ratio)