On 24/01/2013 8:52 a.m., Alex Rousskov wrote:
Hello,I noticed that "tcp://host:port" logger does not get "rotate logs" and other commands that the "daemon:foo" helper gets. The TCP logger just gets the logging line, without any prefix to accommodate future commands or extensions. This seems rather unfortunate because it prevents Squid from reporting significant logging-related events to the remote TCP logger. Log rotation is one example. Another example would be reporting the number of lost (due to connectivity problems) access.log records to a remote logger (if we start supporting such a feature). Is it to late to adjust documentation so that all logging helpers can get logging commands? If it is too late, how about adding an "understands commands" option to access_log so that loggers that can understand commands can receive them? Thank you, Alex.
There are at least two users making use of the TCP output method that I am aware of.
At this point it woudl be a little messy but still possibly worth doing as a concept.
However, this tcp: module is designed as a stream replacement of access.log stdio output all the tools are designed to operate with. Details such as when rotation is requested by the (now external) admin or when Squid starts/stops/reconfigures have never been logged to the access.log files before (udp: and syslog: do not get them either) and cache.log is not sent over the tcp: module.
As for lines lost. That might be worth recording. I am thinking a log line " ... lost NNN lines during TCP connection outage ..." would be sufficient ??
Amos
