Desktop app for analysing BMS kibana logs
If you have no raw logs saved you will need to connect to kibana and pull the latest records.
Activate the ho-it-live VPN before attempting to connect and include an .env file in the root of the repository with your LDAP credentials:
LDAP_USERNAME=...
LDAP_PASSWORD=...
Coordinate processing unfortunately takes a long time due to limitations in the current projection implementation (On^3). This could be improved in the future if needed but in the mean time it is advisable to avoid regenerating large datasets whenever possible.
Run make errors. This will pull all the kibana logs with message types in the known error list from the last month, then compute the similarity between their errorMessage properties.
Run make alerts. This will pull all the kibana watcher executions from the last month that resulted in a successful fire, attempt to locate their associated log, then compute the similarity between the errorMessage properties of all these associated logs. Unfortunately, this is not all that useful, because many executions don't appear to show up in the slack channel at all while others appear in the channel but have duplicate executions.
Once you have generated the data, or sourced pre-generated files, run make dev to start the application in dev mode. Select your generated coordinates file using the file input to see the graph.