netdata
[Feat]: Display banner warning and chart message if a node has an issue with its clock
Problem
When a node has an incorrect clock set this may cause issues when retrieving data for charts (to confirm if this has more impacts). The issues happen when from Netdata Cloud requests are triggered using relative timestamps.
Description
This is an example of an issue reported by a user (discord thread) where the Agent clock wasn't correct and no data was being displayed on Cloud
After the node's clock was fixed data was being displayed correctly on Netdata Cloud.
If on Netdata Cloud we are able to identify to the users that their node(s) have issues with the clock it can help identify why some data isn't displayed. The charts impacted by this should also flag this to the user in some way.
Importance
really want
Value proposition
- Nodes with an issue on the clock will be identified on Netdata Cloud
- Charts will be able to present to the user that no data is visible due to some issue on the node's clock
Proposed implementation
UI/UX:
- For the identified nodes with issues on the clock we could have a top banner appearing on Netdata Cloud saying that X nodes have an issue with their clocks and this may have impacts on retrieving data for the charts (banner similar to the one used for the new architecture migration request to upgrade nodes)
- On the charts we should be able to display that
no datais presented for the period, even though the had replied with some data
Agent/Netdata Cloud:
- Agent should publish some info on NodeInfo message that would allow the check from Netdata Cloud side
- Netdata Cloud BE should check, when NodeInfo messages are processed, if node has some issue with the clock
based on an issue @MrZammler helped troubleshoot the idea to identify these issues directly in Netdata Cloud and warn the users about them come up
@amalkov @MrZammler @papazach @novykh the ticket describes the overall idea, we need to iron out details, but please check and give some feedback
Hi @hugovalente-pm !
As a first step, there is an indication in UpdateNodeInfo message, a field called updated_at. This is the current time when the message is sent, it could might be used to deduce a diff in time between agent/cloud.
