6.10 Meanings of log files

The Cache logs information to four files, the locations for the first three of which are specified from the cached.conf file. (The final file is always stored in the file log in the cache's data directory):

1. cache_access_log: this is Cache access logfile, which logs the hits and misses. The format of this file is one line per access, with the following fields:
   1. timestamp (in RFC 850 date format)
   2. retrieved URL
   3. client's IP address
   4. object size (only non-zero when there's a hit)
   5. hit/miss indicator. This field encodes the access method (TCP, UDP, and our binary protocol), HIT/MISS, and occasionally a reason (e.g., TCP_MISS_TTL means it was a miss because the object had timed out). Note: prior to Harvest Version 1.2, the client's IP address was not logged to this file.

   The cache_access_log also can be written in the httpd common logfile format which is used by many HTTP servers. To enable, set emulate_httpd_log in your cached.conf file to on. By default, the Cache will write this log file in the httpd common logfile format.

2. cache_log: this file contains logs of error messages, such as errors binding to sockets that are already in use (e.g., because you restarted the cache too quickly; see Section 6.11).

3. cache_hierarchy_log: this file logs the parent/neighbor relationships for each hierarchical resolution that occurs (e.g., if an object is faulted from the parent cache, this file will indicate that fact).

4. cache_dir/log: The cache stores objects on disk as numbered files inside of numbered directories. The cache_dir/log file stores the mapping between URL and numbered file.

Note that all logs are appended (rather than recreated whenever the cache starts up).