Details
-
Type:
Epic
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: squash
-
Labels:None
-
Epic Name:sqre-f18-metrics-ui
-
Story Points:72
-
WBS:1.02C.10.02
-
Team:SQuaRE
-
Cycle:Fall 2018
Description
We are concerned that the custom user interfaces we have been making with bokeh for the SQuaSH UI will not scale to the demands for interactions with EFD data. We will evaluate the following off-the-shelf harnesses for the "monitoring and alerts" system described in DMTN-082
- honeycomb.io
- prometheus
- InfluxDB
We are then going to stand up one of those systems as a prototype for first, SQuaSH-DB data and then EFD-originating data.
Attachments
| Key | Summary | Story Points | Assignee | Status | |
|---|---|---|---|---|---|
|
|
|
1.4 | Angelo Fausti | Done | |
|
|
|
Implement a celery task in the SQuaSH API to save time series data to InfluxDB |
5.6 | Angelo Fausti | Done |
|
|
|
1.4 | Angelo Fausti | Done | |
|
|
|
Kubernetes deployment of the InfluxDB + Chronograf + Kapacitor stack |
7 | Angelo Fausti | Done |
|
|
|
4.2 | Angelo Fausti | Done | |
|
|
|
5.6 | Angelo Fausti | Done | |
|
|
|
5.6 | Angelo Fausti | Done | |
|
|
|
5.6 | Angelo Fausti | Done | |
|
|
|
1.4 | Angelo Fausti | Done | |
|
|
|
4.2 | Angelo Fausti | Done | |
|
|
|
0.5 | Angelo Fausti | Done | |
|
|
|
1.4 | Angelo Fausti | Done | |
|
|
|
3.2 | Angelo Fausti | Done | |
|
|
|
0.25 | Angelo Fausti | Done | |
|
|
|
Deploy SQuaSH API to production and update InfluxDB with validate_drp results |
4.2 | Angelo Fausti | Done |
|
|
|
2.8 | Angelo Fausti | Done |
This epic included test deployments of three metric and visualisation environments with the purpose of replacing the SQuaSH front end experience and also with the eye of supporting commissioning usecases requiring low-latency access to EFD data from the Science Platform. Ultimately we selected and deployed the InfluxDB+chronograf+kapacitor chain for scalar metric visualisation as it offered the best ecosystem of applications for our usecases. However we also had a good experience with the prometheus environment and we will be using it for system monitoring in situations where it is better integrated with the monitored infrastructure (eg k8s services).