fluid-cloudnative
[FEATURES] Add the switch parameter of whether to load metadata.
What feature you'd like to add: When starting a cache cluster by creating alluxio runtime, users can choose whether to load metadata. Why is this feature needed: When working with large datasets, the load datameta operation consumes CPU and memory for a long time. In addition, subsequent distributedLoad commands will also load metadata, so the initial load metadata operation is redundant for more usage scenarios.
Hi, @LKR1996 I'm currently working on this. Loading metadata is actually a prerequisite for Fluid to get Dataset summary(e.g. Ufs total size, cached percentage). If we disable loading metadata, users are not able to know such information. Is this okay for your scenario?
Hi, @LKR1996 I'm currently working on this. Loading metadata is actually a prerequisite for Fluid to get Dataset summary(e.g. Ufs total size, cached percentage). If we disable loading metadata, users are not able to know such information. Is this okay for your scenario? Hi, @TrafalgarZZZ Thanks for reply! Dataset summary is not necessary for our scenarios. We just want to get the pv and pvc with the same name as the dataset through fluid for use by the application pod, and we want to load metadata at the right time.