We have a server running a standard LAMP stack with WordPress. It largely woks fine. I use a custom format string in my Apache config that includes the time taken to process the request at the final field in the access logs.
Looking at the access logs I see quite a large number of 408 responses (which should indicate a timeout). However, they have no request line and these "timeouts" are generally only for a few tens of microseconds!
Looking at the current log (last 13 hours) I see over 1500 entries with a 408 responses. One had a duration of 2937771 but none of the others had over 200 microseconds! The only one with a sensible duration is also the only one with a request line, so that probably is a real timeout. The others make no sense to me.
Question: Why are some requests being timed out so quickly while other work? Should I be worried or just ignore them
To give a bit more detail, my httpd conf includes the following format directive (and the site-specific config uses "common" for the access log)
LogFormat "%h %l %u %t \"%r\" %>s %b %D" common
As per the Apache docs, the %D at the end is
The time taken to serve the request, in microseconds.
An example extract from the log, with the source IPs obfuscated looks like
I've found other issues about 408 which generally talk about how to configure sensible timeouts (I haven't configured anything so have the standard apache defaults for this). But haven't found any where they had access to the response time, so my main question is why/how they timeout so quickly.