Consequences of API server time drifting significantly from user's browser time
This ticket tracks our holistic approach to the general problem for which issue #767 is a specific case. A fix for #767 is necessary but not sufficient to fix this issue.
Problem/Opportunity Statement
Clocks can drift away from the correct time on one or more OpenStack API servers, and on an Exosphere user's device. This may cause unexpected / undesired behavior in Exosphere. For example, some of our logic about ExoSetupStatus depends on the difference between when OpenStack says that resource was created and when the current time according to the browser. There are likely other effects too.
What would success / a fix look like?
- Consider holistic / app-wide consequences of the API server time being significantly drifted (more than a few seconds) from the user's browser time
- Decide and implement an overall approach to detecting and dealing with this problem when it happens
- Suggestion: we can decide that synced clocks (under some maximum threshold that we decide on) are a requirement for Exosphere to work well. If Exosphere detects that the time drift is more than X (where X is perhaps a minute), perhaps we simply warn the user of this situation in a top banner, tell them to double-check their computer's clock and/or ask their cloud administrator.
Edited by Chris Martin