You may try turning on slow logs that record more detailed individual milestones (time taken for each internal state during the transaction lifecycle) for transactions that take longer than a specified duration.
https://docs.trafficserver.apache.org/en/latest/admin-guide/files/records.config.en.html#proxy-config-http-slow-log-threshold - Sudheer > On Nov 21, 2016, at 3:19 AM, Mateusz Zajakala <[email protected]> wrote: > > Hi, > > I'm trying to debug a reverse proxy which occasionally takes longer than > usual time to serve a client. I can see in the logs e.g. that the request > took 10 times the usual amount of time for similar file, however I can't say > whether it was caused by the origin side or client side (i.e. was the issue > when retrieving content from origin or serving it to client). > > Is there any way to get more detailed logs for response time? I'm looking at > 'ttms' value, is there any log to indicate "origin connection time", "time > between sending request to origin and finishing reading response", "time > between starting and finishing serving response to client" etc. That would > help with optimizing... > > If there is nothing in the logs, maybe there is some useful statistic I > should look into? > > Thanks > Mateusz
