Usually when I look at the access logs, the timestamps are all in chronological order.
I'm troubleshooting a problem in my app that happens daily, each morning at around 8:15am, so I'm looking at the logs for around that time. I notice that the times aren't sequential. eg:
1.1.1.1 - - [18/Oct/2010:08:16:14 -0400] "GET ..."
1.1.1.1 - - [18/Oct/2010:08:15:14 -0400] "GET ..."
1.1.1.1 - - [18/Oct/2010:08:15:14 -0400] "GET ..."
1.1.1.1 - - [18/Oct/2010:08:15:14 -0400] "GET ..."
1.1.1.1 - - [18/Oct/2010:08:15:14 -0400] "GET ..."
1.1.1.1 - - [18/Oct/2010:08:15:14 -0400] "GET ..."
1.1.1.1 - - [18/Oct/2010:08:16:15 -0400] "GET ..."
1.1.1.1 - - [18/Oct/2010:08:16:15 -0400] "GET ..."
1.1.1.1 - - [18/Oct/2010:08:16:15 -0400] "GET ..."
1.1.1.1 - - [18/Oct/2010:08:15:15 -0400] "GET ..."
1.1.1.1 - - [18/Oct/2010:08:15:15 -0400] "GET ..."
(IP and request changed/omitted for privacy)
See how there are some entries that are timestamped a minute out of sync from those around them?
These are just a few examples - the same asyncronicity is happening fairly often in the logs.
What would cause that?
If the timestamp is a record of when the request was completed and the response sent, then how could they get out of order in the logs? Not just by a second, but by an entire minute? And if the timestamp is recording some other step in the HTTP process, then what does this mean?