* using goreleaser for semver builds on sloop. Integrated with travis ci and docker.
* fix: deploy script update to ignore new lines
* fix: deploy script update to ignore new lines
* fix: using make target instead of shell script as travis having issues with new lines
* fix: removed travis environment variable from makefile
* fix: updated build script
* Added guidance for memory consuption
* Added metric for dropPrefix latency so that we can measure for how much time it locks the DB
* fixed the comment
Upgraded Badger version.
Added flattening at start up time.
Fixed the event count spreading issue which resulted in uneven data distribution across partitions.
Moved to drop prefix as it yields better space claim.
Added feature flag for switching to delete prefix. Also changed the numberfversions to 0 so that delete prefix would reclaim space. dgraph-io/badger#1228
Fixed the issue of unclaimed !badger!move prefixes which are never cleaned up. Details: dgraph-io/badger#1288
Added support in debugging pages to see internal keys.
* Garbage collection fix
Issues fixed:
1. Garbage collection was not able to sustain the required size limit.
2. Exposed more badger options to help in compression
3. Fixed runtime bug encountered when event records with nil map was retrieved.
* Incorporated comments
* Incorporated Duke's comments
* GC Fix
* Addressed all comments and added unit test
Fixed the replay use case and added unit test
* Removed the GetMin and GetMaxPartitions functions as they are not required now
* Added more unit test cases
* Badger Tweaks for memory utilization
* GC Fix
* Addressed all comments and added unit test
Fixed the replay use case and added unit test
* Removed the GetMin and GetMaxPartitions functions as they are not required now
* Added more unit test cases
* Removed dropPrefix to fix the performance
* Added some more metrics
* Added unit tests and some more metrics
* Removed dropPrefix to fix the performance
* Added some more metrics
* Added unit tests and some more metrics
* Incorporated some comments
* Incorporated all the comments
* fixed the minor comment
Solved Problem: Disk usage in our cluster grew way beyond the config limit resulting the application to be OOM killed. There were a couple of issues identified and fixed as follows:
1. Event count table data (although it is the smallest table in terms of number of keys and size) was getting added beyond the max look back time.
2. The sprinkling of event count data was using max look back as a boundary resulting in small partitions containing very less event count data spread all across till look back time.
3. Garbage collection was never called for watch activity table.
4. Added prometheus Metrics for better visibility.
* Split MakeKubernetesClient into Client & Context as to avoid -disable-kube-watch from requiring a valid context
Fixes https://github.com/salesforce/sloop/issues/58
* Put back kubeContext into log line
* Add /data/backup endpoint to dump database
* Use tables.Db() instead of raw db
* Add backupHandler doc
* Provide better error message for since parsing issues
* wip
* Show payloads during window time
This PR adds getPreviousKey logic in getPayload logic.
Add front end to show the payloads during the windown time.
* address comments
* fix the unused code
Sridhar Vemula
Philippe scorsolini
sana-jawad
cmeister-sfdc
Thomas Hargrove
mtmn
DuncanSmith1126
Ash Caire
hsiddulugari
Min Wang
Ryan Brainard
Duke Harlan
nurland