Archive for October 2003
We had so many issues to solve around getting things to work in the deployment environment, correcting errors, and verification, that we hardly got started on verifying the cycle 1 stories. Most were not even started yet. Some were done and rejected with some extra work. We’ll continue this on tuesday (monday is a holiday).
See comments in individual stories for some comments.
Micromuse installed some SSM’s and ASM’s today in decafe, for us to play with. The configuration tool is on backup01.
Event flow in Decafe now works end-to-end using the stub helpdesk. We solved a number of problems due to config files being overwritten during the auto deploy.
- Service correlator now does not cache Component state. It does cache dependency information, so that lookup is fast.However, it now is stateless with respect to component states. The old Correlator used a Dataset representation of components and dependencies.
The new implementation reuses the ComponentCache that is in use in reporting.
- Added writeCount protocol, so that multiple highly available service correlatorscan process the same event without interference.
- Added NUnit test cases for all of the above. These are not added to Nant yet, in order not to brake the build on the buildmachine.These testcases create a new database, and assume the presence of SQL and a certain SA password.
All alerts in Netcool are stuck in status 2 (raised to Smart24). This means they never saw a ticket-opened or closed notification coming back into Netcool. I investigated a bit, and found that the SocketGatewayListener, which is a perl script that is listening for messages from Netcool’s socket gateway, is not receiving any messages. The listener itself is working fine, since I can telnet into it and send a message manually. However, the socket gateway appears to be working OK too. I can see in its log that it is picking up alerts and sending them (internally) to its writer.
This is a mystery to me. Perhaps somebody can shed some light on this
The CVS sourceforge problems is fixed. The tracker / commit log integration has been reactivated.
When you commit in CVS, you will see a “tracker:” text. If you enter the number of the story you are working on here, links will be created automatically from the story to the checked-in code in Sourceforge
This changes the CVS config, so you need to check-out anew in a virgin directory. I sent an e-mail on this this afternoon. Unfortunately it didn’t get delivered due to a problem wiht our Exchange server. I resent the message tonight.
- Database creation / deletion scripts for Service correlator test cases
- Created database read test cases for Service Correlator.
- Partly implemented new Service correlator cache until read testcase succeeded
- Fixed Exchange name lookup problem (by rebooting!)