HDDS-15664. ozone debug ldb display DeletedBlocksTransactionSummary info

[ChenSammi]

What changes were proposed in this pull request?

add support in ldb scan command, to display DeletedBlocksTransactionSummary info.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-15664

How was this patch tested?

Manual test

root@vc1337:/var/log/hadoop-ozone# ozone admin scm deletedBlocksTxn summary
RemoteException: SCM Server, while invoking StorageContainerLocationProtocolPB over node1. Retrying in 2000ms after 0 failover attempt(s).
DeletedBlocksTransaction summary:
  Total number of transactions: 43
  Total number of blocks: 43
  Total size of blocks: 2709
  Total replicated size of blocks: 8127


root@vc1337:/var/log/hadoop-ozone# ozone debug ldb --db=/var/lib/hadoop-ozone/scm/data/scm.db/ scan --cf=statefulServiceConfig
{ "DeletedBlockLogStateManager": {
  "memoizedHashCode" : 0,
  "memoizedSize" : -1,
  "unknownFields" : {
    "fields" : { }
  },
  "bitField0_" : 15,
  "totalTransactionCount_" : 43,
  "totalBlockCount_" : 43,
  "totalBlockSize_" : 2709,
  "totalBlockReplicatedSize_" : 8127,
  "memoizedIsInitialized" : 1
}
 }

[priyeshkaratha]

Thanks @ChenSammi for the patch. Changes LGTM

[Tejaskriya]

Thanks @ChenSammi, LGTM!

[adoroszlai]

Thanks @ChenSammi for the patch. DBScanner is very generic, but the PR adds tightly coupled parse logic. Can you please explain why this is needed, why STATEFUL_SERVICE_CONFIG is the only column family that needs special care? If the current logic cannot parse it, can it be implemented in a way that helps with any other column families not handled currently?

[adoroszlai]

Duplicate catch blocks, please merge them.

[ChenSammi]

@adoroszlai , STATEFUL_SERVICE_CONFIG is a table which doesn't belong to a specific feature of SCM, it's used as a key-value store(String -> ByteString) for all SCM features to persist value which need persistent. Currently it's used by container balancer, and Root CA rotation manager, now storage capacity feature. So the element in this table is specific to a feature, not like other tables in SCM, which has the same value data structure.
Currently, DBScanner can display this table with it raw ByteString value, not a understandable format of human, equal to no use during the debug time.
This task aims to covert the ByteString value to a storage capacity feature data structure and display it, so DBScanner will be helpful in the debug of future storage capacity feature. I'm not sure if I explain it clearly enough?

[adoroszlai]

Thanks @ChenSammi for the explanation.

I would like to suggest some refactoring to reduce hard-coding in DBScanner, and also support container balancer and Root CA rotation manager: adoroszlai@0d5dd70 I can also submit follow-up PR if you prefer.

patch updated