Siebel: A Simple Log Analyser

 

Deciphering Siebel log files is a chore. The basic premise has not varied since a long time now:

1. Look out for all the errors logged in the file. Even if most of them are benign, there is a chance (bad word to use here) that they are contributing to a larger error/system crash
2. Keep a look out for the performance of various SQLs thrown at DB. If a transaction is taking a long time, that’s where you want to start
3. Verify whether any Siebel transaction outside the DB is taking more time than expected (e.g. in cases where you have a really complex search spec that cannot be passed on to the DB)

Although I wanted to automate this somehow, for a very long time I just relied back on “proven” methods:

• Open log file, search for execution times around the functionality that is causing the issue
• Search for errors – most likely around the section that is causing the issue
• Navigate back and forth a lot, so that rest of the people think I already have a feel for how Siebel works and doing something great
• Rely on grep, tail and perl (when working on UNIX and developer-friendly). Sadly, Windows does not provide it out of the box, and to get open source/free tools in a restricted environment is no joy

This changed when I took a look back at  where .NET 2010 is, the culmination of which is here. “COGSiebelLogaliser” is a tool to parse Siebel logs generated by a client or server, and spool files to display useful information about:

• Performance times
• Errors
• An included editor is used to open log files (faster to open and navigate through large files)

High level requirements:

1. Tested it in Windows XP, with .NET 4.0
2. Application should have write access to the folder from which it executes, and read access for the log files. It will not write back anything to the log files

Of course this has to be treated as Alpha version since it has gone through limited testing. If you have blind trust in executables supplied from any person, and courageous enough to test out new things, you can try the application here. The tool must be tested against sample log files in a local, isolated environment until you are sure what it does, and how it works. Source code is pretty amateurish, and has to wait for a later date to get released to the wild.