Present
- Williamson, Rob
- Morrisey, Scott
- Gutta, Srinva
- Brown, Leo
Discussion
- camunda application yaml
- camunda auto-deployment config
- rough outline of auto-deployment shared
- Scan java classpath for .bpmn files (might be .xml as well)
- Parse each bpmn
- For each process id in each file
- Select on ACT_RE_PROCDEF to see if the process is present
- Compare db and in memory hash to determine if deployment is necessary
- {if hash is diff} INSERT
- rough outline of auto-deployment shared
- apollo-camunda-orchestrator inc. bpmn
- apollo docker config
- cct-camunda-orchestrator inc. bpmn
- new relic of specific apollo orchestrator pod
- recorded memory and cpu requests (k8s and jvm config)
- k8s report
Actions
- Backlog tickets to review Readiness / Liveness probe timeouts to be in line with the app requirements - need domain understanding
- Increase CPU Requests for Camunda and possibly other CPU hungry services to distribute better??
- Remove CPU limits from all services
- Check Java XMX and XMS are in line with k8s limits
- Find out what the defaults
- Camunda turn off auto deployment of processes – and build in a process to update BPMN on the deployments
- test on Thursday