> @Mark S > Logging is something I've been very interested in at various points in > the past, usually when an application goes wrong in production and I > have to figure out what went wrong! > > I think we need to discard the notion that logs should be timestamped
I still like timestamps even with per-user logging. Since each user has its ownfile yyyy-mm-dd/userid the timestamp per line only needs to contain HH:MM:SS. Each day a new directory/file is created as needed. So log files don't grow toolarge. Another thread removes dirs/files older than 30 days. > strings outputted to a file. Instead, I think they should be hash-maps > we send to a logging server, which then stores them in a database. I prefer files. Using a DB prevents me from using existing powerful tools to search and query the logs: vim and emacs. The nice thing about per-user logging files is that they don't get that big. They are quick to load/search/analyze. > You could then visit the database and say "Show me all exceptions that > occurred for user X". You could then select the field you wanted and > say "Show me all log messages that occurred within the context of the > same request as this exception". This is something that would have > saved me a lot of pain in the past. Wrt context: per-user logging data is written serially so it's easy to figure out the context/progression of events that lead up to the exception. You know where one request starts and ends. FYI: we use the STM to queue up requests in a map and flush each user's loggingdata to disk every few seconds. -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en
