collect metrics in a separate go routine #82
Conversation
marten-seemann
force-pushed
the
fix-unbounded-state
branch
from
79b9740 to
544cba2
Compare
metrics.go
Outdated
| info, err := conn.getTCPInfo() | ||
| if err != nil { | ||
| if strings.Contains(err.Error(), "use of closed network connection") { | ||
| c.closedConn(conn) | ||
| continue | ||
| } | ||
| log.Errorf("Failed to get TCP info: %s", err) | ||
| continue | ||
| } |
There was a problem hiding this comment.
Is it a problem that if we conn.getTCPInfo() errors differently from expected such that we'll never close the connection?
I've seen 2021-07-08T21:39:47.925-0400 ERROR tcp-tpt [email protected]/metrics.go:118 Failed to get TCP info: raw-control tcp 192.168.1.6:4001: getsockopt: not implemented on Windows so I'm concerned this could cause us problems.
There was a problem hiding this comment.
I tried it in my Windows VM, and I'm getting the same error. Should we disable metrics collection on Windows?
|
Following up here, with the recent Slack conversations in mind.
That is the Prometheus idiomatic way. In other words, just-in-time (on scrape / on |
marten-seemann
force-pushed
the
fix-unbounded-state
branch
from
2851bff to
43b076c
Compare
|
@marten-seemann : what are the next steps here? Who do we need a review from? |
marten-seemann
force-pushed
the
fix-unbounded-state
branch
from
43b076c to
660cdd6
Compare
marten-seemann
force-pushed
the
fix-unbounded-state
branch
from
660cdd6 to
d611dc7
Compare
I wasn't done yet with my changes, that's why I didn't request any reviews. Now I am, and will do so. |
|
@aschmahmann @Stebalien I updated the PR. Now
|
Fixes #81.
I really wish there was a way to pass a Prometheus struct to tcp-transport, and only collect metrics in case metrics are actually scraped. This PR introduces an eternal Go routine that periodically asks the kernel for metrics, and removes a connection from connection map when it is closed.