fabio
fabio copied to clipboard
Excessive TLS handshake error logging
I am behind an Amazon ELB which gets probed with a lot of non-TLS traffic which fabio is logging as
invalid log msg: 2017/12/12 20:21:35 http: TLS handshake error from 10.65.241.135:19568: EOF
Note the "invalid log msg", I'd like to be able squash these messages by setting the log level to INFO or something but I don't think they are being logged properly to be filtered. It would be even more useful if the source ip address (upstream of the ELB) could be logged here in case this is misconfigured internal traffic (unlikely).
This new comes from https://github.com/fabiolb/fabio/blob/master/logger/level_writer.go#L35 which is probably triggered by a log.Print
without a prefix. I’ll grep through the code later.
Is there any fix available for this issue?
Commenting up on this one, our stderr logs are filled with this constantly. We do have a TCP health check on the https port, and this also triggers it constantly.
This looks to be coming from the net.http package: https://github.com/golang/go/blob/master/src/net/http/server.go#L1826
Setting the http.Server.ErrorLog to something non-nil will allow capture of this error, otherwise it goes to log.Printf.
Is there any fix available for this issue?