Now I am using gradle mlReloadSchemas tasks to reload TDE templates. I guess even if the change is for one tde file only, the reload schemas task may delete all in DB and load all TDE templates to ML DB.
I wonder whether it will cause a performance issue for ML. Will that trigger indexing even for the TDE files that have not yet changed?
I am using DevOps pipeline to trigger the schema reload from GIT repository. As such, I could not load only the change TDE file. I have to reload everything. If there is performance issue, how to load only changed file with the pipeline?
CodePudding user response：
Redeploying TDE can cause reindexing. How many records to be reindexed depends upon the context matching for those TDE.
A properly resourced cluster should be able to handle the load of reindexing.
That being said, the merging activities can compete with online traffic and query demands. You can help minimize the impact by setting the reindex throttle to a lower level (1-5 with 1 being the lowest), and you can set a background-io limit to restrict the amount of IO any node will use for background activities such as merges and backups.
You can also choose when to enable/disable reindexing, and adjust the reindexing level to a higher/lower level at different periods.