- added filter and stats icons to log stream labels
- removed click handler from label itself
- click on stats icon calculates label value distribution across loaded logs lines
- show stats in hover
- stats have indicator which value is the current one
- showing top 5 values for the given label
- if selected value is not among top 5, it is added
- summing up remaining label value distribution as Other
- Logging datasource sends error string directly in the message body
- allowing response body to be an error when query transaction failed
- added throw in case we ever encounter unhandled errors again (forces us to fix the issue)
- ensure that selectors have to be preceded by whitespace or line beginning
- glacing over the fact that ` {0,1}` could be a valid regexp, but we're using space as query/regexp separators anyway, so I think it's fine for now
- fork promql's tokenizer (need to specify that labels context can only follow beginning of line or whitespace)
- remove unneeded syntax features
- only present history items when field is empty
- fixed typo
- moved result calculation to explore utils
- use component keys instead of componentWillReceiveProps
- require logs to have and id, to as use as Logs component key
- render delay based on row count
- moved from grid to flexbox
- calculate Explore results only when query transactions change to prevent expensive re-renders
- split up rendering of graph and log data
- render log results in 2 stages
- adds a custom label renderer to Logs viewer in Explore
- labels are no longer treated as strings, they are passed as parsed objects to the log row
- label renderer supports onClick handler for an action
- renamed Explore's `onClickTableCell` to `onClickLabel` and wired up log label renderers
- reuse Prometheus `addLabelToSelector` to modify Logging queries via click on label
- added tests to `addLabelToSelector`, changed to include the surrounding `{}`
- use label render also for common labels in the controls panel
- logging meta data section has now a custom renderer that can render numbers, strings, and labels
- style adjustments
Sometimes log streams dont return any lines for the given range. Would be great to automate the search until some logs are found.
- Allow Explore to drive TimePicker via ref
- Show `Scan` link in Logs when there is no data
- Click on `Scan` sets Explore into scanning state
- While scanning, tell Timepicker to shift left
- TimePicker change triggers new queries with shifted time range
- Remember if query transaction was started via scan
- keep scanning until something was found
- Manual use of timepicker cancels scanning
- since the DataQuery rewrite, Explore starts by submitting its queries (Explore cant know if the queries are emtpty)
- The datasource intercepts an empty query and returns an empty list of time series, and also no table for the table query.
- The query hinter then received no series to analyse and was not guarding against this.
- This PR adds this guard.
- add `onToggleSeries` to Explore Graph props
- toggling a Logging Graph series from the legend propagates its hidden series to Logs
- Logs translates hidden series alias to LogLevels
- Logs filters out hidden log levels
Logging's query importer drops all labels that are not common to both datasources.
- keep labels if label lookup of logging datasource came back empty
- easier to remove labels than having to add them in the first place
Currently the footer is based on top margins which looks odd on big screen sizes.
Also, the footer shows up on the top of the page on pages take a moment to load (while ng-view is rendering).
- True sticky footer based on flexbox
- Ensure footer stays at bottom while ng-view div does not exist yet
- Queries in Explore have been string based
- This PR introduces the use of the DataQuery type to denote all queries handled in Explore
- Within Explore all handling of DataQueries is transparent
- Modifying DataQueries is left to the datasource
- Using `target` as variable names for DataQueries to be consistent with the rest of Grafana
- added dedup switches to logs view
- strategy 'exact' matches rows that are exact (except for dates)
- strategy 'numbers' strips all numbers
- strategy 'signature' strips all letters and numbers to that only whitespace and punctuation remains
- added duplication indicator next to log level
- replace the Graph/Table buttons with toggle control in a wrapper panel
- moved toggle control to left to be close to the label
- removed panel styles from Logs and Graph viewer
- moved loader animation to panel
Rules expansion (available via query hints in explore) was broken for expressions that contained selectors.
- fix replacing regexp to recognize `{` and `[` as the end of a rule name
- moved logic to language utils
- added tests
Explore is about keeping context between datasources if possible. When
changing from metrics to logging, some of the filtering can be kept to
narrow down logging streams relevant to the metrics.
- adds `importQueries` function in language providers
- query import dependent on origin datasource
- implemented prometheus-to-logging import: keeping label selectors
that are common to both datasources
- added types
Datasource selection triggers a connection test, on success
the DS is set in the Explore state. If the test takes long and user
selects a different DS, and just after that the first test succeeds,
then the first DS overwrites the state.
* when test returns check if datasource is still the requested one
Tab completion gets in the way when constructing a query from the inside
out:
```
up| => |up => sum(|up)
```
At that point the language provider will not suggest anything.
- Logging gets a graph for log distribution (currently per stream, but I
think I'll change that to per log-level)
- added grid columns for timestamp and unique labels
- show common labels of streams
- View options to show/hide time columns, label columns
- created `--small` modifier for Switch CSS classes
- merging of streams is now a datasource responsibility
- changed `start()` to return promise on main language feature task
- promise resolves to list of secondary tasks
- speeds up time to interaction of metric selector
- lazy loading of certain metric selector and log label selector items
- loading indication of metric and log label selectors
this is a fix-up PR that cleans up Explore Logging after the recent
restructuring.
- log results need to be merged since query transactions have been
introduced
- logging DS has its own language provider, query field, and start page
(some of them based on prometheus components)
- added loader animation to log viewer
- removed logging logic from prometheus components
- using grunt-newer to prefix precommit tasks
- only got it to work for tslint and tsc
Not applied to:
- sasslint does not take the file arguments in a way that grunt-newer
recognizes
- no-only-tests throws an error when used with `newer`, but it's
sub-second runtime
- when instantiating a datasource, the datasource service checks if the
plugin module exports Explore components, and if so, attaches them to
the datasource
- Explore component makes all major internal pluggable from a datasource
`exploreComponents` property
- Moved Prometheus query field to promehteus datasource and registered
it as an exported Explore component
- Added new Start page for Explore, also exported from the datasource
- added `node_modules` as new target
- dependency on `package.json` and `yarn.lock` allows for quick `make
node_modules` after a branch change, which noops when the deps have
not changed
- also added `clean` target
- use global range types
- add ErrorBoundary around individual Explore components
- fix table merge on empty results
- fix TimePicker date parsing on ISO dates
- fix TimePicker range string after relative move
David