when we made image upload optional we didnt
show the default value properly in the UI. Which
caused confusing. This commit apply the default
values to existing notifiers in the edit pages
and reverts back to using uploadimage=true as the
default value.
This changes the logic for the alert validation in the extractor. Validation
is executed twice, once before attempting to save the alerts and once after
saving the dashboard while saving the alerts to the db. The first validation
will not require that the alert has a dashboard id which allows new dashboards
with alerts to be saved.
Fixes#11247
Use the ConvertSqlTimeColumnToEpochMs function to convert any native
datetime data type or epoch time (millisecond precision).
Refactored mysql implementation to make it more similar to postgres
and mssql implementations.
Added $__timeEpoch macro function with same implementation as $__time.
Added possibility to use a time column named time in addition to
the currectly supported time_sec.
Additional tests and update of existing.
Added test dashboard.
Use the ConvertSqlTimeColumnToEpochMs function to convert any native
datetime data type or epoch time (millisecond precision).
Additional tests and update of existing due to timezone issues
running postgres on UTC and dev environment on non-utc.
Added test dashboard.
Use the ConvertSqlTimeColumnToEpochMs function to convert any native
datetime data type or epoch time (millisecond precision).
Additional tests and update of existing due to timezone issues
running MSSQL on UTC and dev environment on non-utc.
Update stored procedures test to handle more parameters.
Update test dashboard.
Removes the macro function . Macro functions should not do UTC/timezone conversion - they should
work in the same way as postgres and mysql datasource implementations.
Grafana and Microsft SQL Server should be run on servers with UTC timezones.
Earlier the division of interval was done using whole numbers resulting in that important information
was lost/too many time series merged to the same group. Now using division of floating point and rounding
up to solve the problem
This seems like a niche feature which can be solved by defining multiple queries. In the future
we'll probably add support for defining series name by alias field similar to how other datasources
have solved that, e.g. prometheus.
