GeorgeJahad

Hi @tim-finnigan: Just wondering if anyone has had time to consider this?

What is the plan for synchronization within the SnapshotChain structures? Snapshot diff could try to read at anytime, right?

A flaky test, unrelated to snapshotting, was preventing the tests from passing since I did the rebase. I debugged it and added the fix to this pr, (just so I...

Thanks for asking, @adoroszlai , but I'm too busy to look at this right now.

There is a call to looupKey here: https://github.com/apache/ozone/blob/1ea4c2adc640a02508696bf346c7f1f7b6e0afd9/hadoop-ozone/client/src/main/java/org/apache/hadoop/ozone/client/rpc/RpcClient.java#L1264 Should that be changed to getKeyInfo()?

There is a call to lookupFile() here: https://github.com/apache/ozone/blob/1ea4c2adc640a02508696bf346c7f1f7b6e0afd9/hadoop-ozone/client/src/main/java/org/apache/hadoop/ozone/client/rpc/RpcClient.java#L1683 But lookupFile() doesn't appear to use the cache. Should it?

So there is a problem with clearing the NSSummary table during reprocess(). Both the fso and legacy tasks write to the nssummary table. The reprocess() method for FSO clears the...

Also because the two tasks are writing to the same table from separate threads, it would be good to clarify why no synchronization is needed.

I just tried running: "ozone admin namespace summary /vol1" It gets a warning: ``` [Warning] Namespace CLI is only designed for FSO mode. Bucket being accessed must be of type...

LGTM