Yup, that is one of my fallback approachs. It seems to me that I should be able to write a configuration for the Zopes that either invokes a remote syslogd server (or other server) which will merge multiple logging entries into a single file on one of the machines in the cluster, preferably the ZEO host. Another approaches would be to maintain local logs and then have a process running on the remote machine that performs the merge. We post-process the logs in nearly real-time and we expect related transactions to spread across multiple Zopes. On Mon, 1 Aug 2005, Andreas Jung wrote:
--On 1. August 2005 09:51:37 -0700 Dennis Allison <allison@shasta.stanford.edu> wrote:
Zope 2.7.6 with ZEO Python 2.4.1
I would like to configure for a single merged log. If BE is the backend machine running ZEO and FEA FEB and FEC are various front-ends for the ZEO, what's the configuration supposed to look like to get a merged log collected on the machine on which ZEO runs. The only log I need to merge is the Z2.log--the othes can be left per-Zope frontend. While it would be nice if the merged entries were serialized, it is not strictly necessary. Cross machine transmission can be either UDP or TCP.
I don't know what the sense of a merged Z2.log is. You can install a front-end proxy + load balancer so you could use the logfile of the proxy or write a small script to merge the the Z2.log from multiple clients into a single file. You should be straight forward with some line of Python.
-aj
--