bokuholio.RemoveThis@gmail.com wrote:
> I would like to find a tool, where from a central server, I can
> collect those daily rolled logs from each of the servers. Ideally it
> would verify that the file was received, remove it from the host, gzip
> it up and retain it for some predetermined period in a directory
> hierarchy.
you've got a lot of options:
REMOTE LOGHOST
someone in this thread already mentioned that you can configure your webservers
to log to a centralized syslog loghost:
http://www.oreilly.com/pub/a/sysadmin/2006/10/12/httpd-syslog.html
REMOTE THIRD PARTY ANALYTICS
if you just need marketing analytics (e.g. unique cookied visitor behavior), not
operational metrics, then you can install third-party analytics on each machine,
and disable the local logging altogether:
http://google.com/analytics/
REMOTE MESSAGING PROTOCOL
if your machines are on the same physical network, you can use the spread
messaging library and mod_log_spread to send all log data to a central recorder:
http://www.backhand.org/mod_log_spread/
https://alioth.debian.org/frs/?group_id=30823
REMOTE DATABASE
if you use a database like postgresql or mysql, you can configure your web
servers to record their log entries to a remote database server:
http://pgfoundry.org/projects/modpglogger
http://bitbrook.de/software/mod_log_mysql/
UTILITIES FOR LOCAL LOG MAINTENANCE
for local log file maintenance, i don't know of anything that eliminates the
scripting "glue" that is necessary to push or pull rotated logs to a central
location, but you can use a few things to make it easier:
cronolog can be configured as an inline logging application that will write out
uniquely named logfiles as often as you need them:
http://cronolog.org/
rsync is a differential / incremental file transfer tool, you can use it to pull
files over to your loghost, and delete remote files after they've been safely
written to the central machine. if a transfer is interrupted, rsync will pick up
where it left off, which can safe you a lot of time:
http://rsync.samba.org/
if you're trying to merge logs from load-balanced web servers into a single
comprehensive log corpus, you can use perl or shell commands to sort and de-dupe
the data, but you'll find this utility to be much faster:
http://mergelog.sourceforge.net/
hth
--sean
--
sean dreilinger -
http://durak.org/sean/ >> Stay informed about: Log Management