Combined log file import fails if user-agent is empty about goatcounter HOT 4 CLOSED

The problem seems to be that the field is "", rather than "-" to indicate an empty value. It scans for .+? (that is, at least one character).

As far as I know, a - should always be used to indicate missing data, but none of this is a "real" standard.

Was this generated by some common(-ish) software?

from goatcounter.

Was this generated by some common(-ish) software?

That's a real access log file of nginx, from a real website. Such records are added when user-agent header is present but empty (user-agent: ).

from goatcounter.

I could reproduce it like this:

$ nc localhost 8090
GET / HTTP/1.1
User-Agent:

HTTP/1.1 400 Bad Request
...

Gave me:

127.0.0.1 - - [17/May/2023:22:04:40 +0200] "GET / HTTP/1.1" 400 157 "-" ""

The trick is to have User-Agent header with no value, rather than User-Agent being omitted (e.g. curl -H 'User-Agent: ' will simply not send any header).

from goatcounter.

Ah, it seems you edited your message. Well, we came to the same conclusion 😅

Mostly just wanted to verify it's not an issue with a custom script or the like. I'll get it fixed.

from goatcounter.