* Explore: propose action, thunk, and decorators for load more
* LogsContainer: add loadMore method
* Query: remove unused var
* Loading more: use navigation to simulate scrolling
* Explore: figure out data combination
* Fix imports
* Explore: deduplicate results when using query splitting
* LogsNavigation: add scroll behavior
* Remove old code
* Scroll: adjust delta value
* Load more: remove refIds from signature
We can resolve them inside Explore state
* Load more: rename to loadMoreLogs
* Infinite scrolling: use scrollElement to listen to scrolling events
* Explore logs: add fixed height to scrollable logs container
* Logs: make logs container the scrolling element
* Logs: remove dynamic logs container size
It works very well with 1 query, but breaks with more than 1 query or when Logs is not the last rendered panel
* Logs navigation: revert changes
* Infinite scroll: create component
* Infinite scroll: refactor and clean up effect
* Infinite scroll: support oldest first scrolling direction
* Infinite scroll: support loading oldest logs in ascending and descending order
* Infinite scroll: use scroll to top from logs navigation
* Logs: make logs container smaller
* Logs: make container smaller
* State: integrate explore's loading states
* Infinite scroll: add loading to effect dependency array
* Infinite scroll: display message when scroll limit is reached
* Infinite scroll: add support to scroll in both directions
* Infinite scroll: capture wheel events for top scroll
* scrollableLogsContainer: deprecate in favor of logsInfiniteScrolling
* Infinite scroll: implement timerange limits
* Infinite scroll: pass timezone
* Fix unused variables and imports
* Infinite scroll: implement timerange limits for absolute time
* Infinite scroll: fix timerange limits for absolute and relative times
* Infinite scroll: reset out-of-bounds message
* Logs: make container taller
* Line limit: use "displayed" instead of "returned" for infinite scrolling
* Infinite scrolling: disable behavior when there is no scroll
* Remove console log
* Infinite scroll: hide limit reached message when using relative time
* Logs: migrate styles to object notation
* Prettier formatting
* LogsModel: fix import order
* Update betterer.results
* Logs: remove exploreScrollableLogsContainer test
* Infinite scroll: display loader
* Infinite scroll: improve wheel handling
* Explore: unify correlations code
* Explore: move new function to helpers
* Remove comment
* Fix imports
* Formatting
* Query: add missing awaits in unit test
* Logs model: add unit test
* Combine frames: move code to feature/logs
* Explore: move getCorrelations call back to query
It was causing a weird test failure
* Fix imports
* Infinite scroll: parametrize scrolling threshold
* Logs: fix overflow css
* Infinite scroll: add basic unit test
* Infinite scroll: add unit test for absolute time ranges
* Formatting
* Explore query: add custom interaction for scrolling
* Query: move correlations before update time
* Fix import in test
* Update comment
* Remove comment
* Remove comment
* Infinite scroll: report interactions from component
* Fix import order
* Rename action
* Infinite scroll: update limit reached message
* Explore logs: remove type assertion
* Update betterer
* Query utils: add function to identify distinct queries
* Chore: return false when the expected node is found
* Query splitting: exclude distinct queries from splitting
* Group queries excluded from splitting
* Update chunking mentions in the code and ui
* Rename files and exported functions
* Rename configuration attribute
* Rename grouped querying function name
* Update more function names
* Update unit test
* Update unit tests
* More renames
* Rename time splitting functions
* Rename splitting files to chunking
* Rename time chunking functions
* Move response functions to response utils
* Remove some blank spaces
* Add an extra test case for frame refIds and names
* add `correlationId` to queries
* trace correlation id in backend
* add correlation id to loki's span
* add correlation id to query chunks
* fix test
* fix DataSourceWithBackend test
* change to `queryGroupId`
* remove empty line
* fix test in `DataSourceWithBackend`
* Range splitting: allow splits larger than 30 days while keeping the fail safe
* Remove limit
* Fix test
* Remove unnecessary try/catch
* Remove unused export
* Query splitting: enable instant queries
* Range splitting: send instant queries as another request group
* Range splitting: increase grouped splitted requests stability
We were defaulting to the `0` index as the first group for the next request batch, but there was no guarantee that the group `0` had a `.partition` entry for `requestN-1`. Now we find the first defined and use that index as the next starting group.
* Range splitting: update unit test
* Range splitting: group metric and logs queries
* Range splitting: intercalate logs and metric queries when > 1 queries
* Range splitting: fix possibly undefined error and remove console log
* Range splitting: update test imports
* Range splitting: add unit tests for multiple queries
* Query splitting: use lodash partition function
* Chore: rename variable
* Chore: attempt to improve readability
* Range splitting: range splitting function
* Range splitting: experiment with 1 hour splits
* Range splitting: reorganize code
* Range splitting: improve code readability and meaning
* Range splitting: add partition limit to prevent infinite loops
* Range splitting: add error handling
* Range splitting: disable for logs queries
* Range splitting: support any arbitrary time splitting + respect original from/to in the partition
* Chore: remove console logs
* Chore: delete unused import
* Range splitting: actually send requests in sequence
* Range splitting: do not split when > 1 query
* Range splitting: combine frames
* Chore: rename function
* split in reverse
* polished reversing
* keep reference to the right frame in the result
* Range splitting: change request state to Streaming
* Range splitting: fix moving only 1 unit of time instead of the provided one
* Chore: change default parameter to timeShift = 1
* Range splitting: do not split for range queqries
* Range splitting: add initial support for log queries
* Range splitting: do not use MutableDataFrame
It has bad performance and it's not required
* Chore: remove unused export
* Query Splitting: move to module
* loki: split: fix off-by-one error (#62966)
loki: split: fix off-by-one loop
* Range splitting: disable for logs volume queries
* Range splitting: combine any number of fields, not just hardcoded 2
* Range splitting: optimize frame-combining function
* Range splitting: further optimize
* Range splitting: combine frame length
* Range splitting: combine stats
* Range splitting: combine stats without assuming the same order
* Query splitting: catch and raise errors
* Range splitting: create feature flag
* Range splitting: implement feature flag
* Range splitting: add unit test for datasource query
* Range splitting: add basic test for runPartitionedQuery
* Range splitting: add unit test for resultLimitReached
* Range splitting: test frame merging
* Chore: fix unit test
---------
Co-authored-by: Sven Grossmann <svennergr@gmail.com>
Co-authored-by: Gábor Farkas <gabor.farkas@gmail.com>
Matias Chomicki
Ivana Huckova
Galen Kistler
Torkel Ödegaard
Sven Grossmann
Domas
Gábor Farkas
Ashley Harrison