[HDFS-15364] Support sort the output according to the number of occurrences of the opcode for StatisticsEditsVisitor

• Type: New Feature
• Status: Resolved
• Priority: Minor
• Resolution: Invalid
• Affects Version/s: 3.2.1
• Component/s: tools
• Assignee: bianqi
• Reporter: bianqi
• Created: 19/May/20 16:08
• Updated: 26/May/20 03:48
• Resolved: 26/May/20 03:48

Description

      At present, when we execute `hdfs oev -p stats -i edits -o edits.stats`, the output format is as follows, and all the opcodes will be output once.

VERSION : -65
OP_ADD ( 0): 2
OP_RENAME_OLD ( 1): 2
OP_DELETE ( 2): 0
OP_MKDIR ( 3): 5
OP_SET_REPLICATION ( 4): 0
OP_DATANODE_ADD ( 5): 0
OP_DATANODE_REMOVE ( 6): 0
OP_SET_PERMISSIONS ( 7): 4
OP_SET_OWNER ( 8): 1
OP_CLOSE ( 9): 2
OP_SET_GENSTAMP_V1 ( 10): 0
OP_SET_NS_QUOTA ( 11): 0
OP_CLEAR_NS_QUOTA ( 12): 0
OP_TIMES ( 13): 0
OP_SET_QUOTA ( 14): 0
OP_RENAME ( 15): 0
OP_CONCAT_DELETE ( 16): 0
OP_SYMLINK ( 17): 0
OP_GET_DELEGATION_TOKEN ( 18): 0
OP_RENEW_DELEGATION_TOKEN ( 19): 0
OP_CANCEL_DELEGATION_TOKEN ( 20): 0
OP_UPDATE_MASTER_KEY ( 21): 0
OP_REASSIGN_LEASE ( 22): 0
OP_END_LOG_SEGMENT ( 23): 1
OP_START_LOG_SEGMENT ( 24): 1
OP_UPDATE_BLOCKS ( 25): 0
OP_CREATE_SNAPSHOT ( 26): 0
OP_DELETE_SNAPSHOT ( 27): 0
OP_RENAME_SNAPSHOT ( 28): 0
OP_ALLOW_SNAPSHOT ( 29): 0
OP_DISALLOW_SNAPSHOT ( 30): 0
OP_SET_GENSTAMP_V2 ( 31): 2
OP_ALLOCATE_BLOCK_ID ( 32): 2
OP_ADD_BLOCK ( 33): 2
OP_ADD_CACHE_DIRECTIVE ( 34): 0
OP_REMOVE_CACHE_DIRECTIVE ( 35): 0
OP_ADD_CACHE_POOL ( 36): 0
OP_MODIFY_CACHE_POOL ( 37): 0
OP_REMOVE_CACHE_POOL ( 38): 0
OP_MODIFY_CACHE_DIRECTIVE ( 39): 0
OP_SET_ACL ( 40): 0
OP_ROLLING_UPGRADE_START ( 41): 0
OP_ROLLING_UPGRADE_FINALIZE ( 42): 0
OP_SET_XATTR ( 43): 0
OP_REMOVE_XATTR ( 44): 0
OP_SET_STORAGE_POLICY ( 45): 0
OP_TRUNCATE ( 46): 0
OP_APPEND ( 47): 0
OP_SET_QUOTA_BY_STORAGETYPE ( 48): 0
OP_ADD_ERASURE_CODING_POLICY ( 49): 0
OP_ENABLE_ERASURE_CODING_POLIC ( 50): 0
OP_DISABLE_ERASURE_CODING_POLI ( 51): 0
OP_REMOVE_ERASURE_CODING_POLIC ( 52): 0
OP_INVALID ( -1): 0

 But in general, the edits file we parse does not involve all the operation codes. If all the operation codes are output, it is unfriendly for the cluster administrator to view the output.
    we usually only care about what opcodes appear in the edits file.We can output the opcodes that appeared in the edits file and sort them.

For example, we can execute the following command：

 hdfs oev -p stats -i edits_0000000000000001321-0000000000000001344 -sort -o edits.stats -v

The output format is as follows：

VERSION : -65
OP_MKDIR ( 3): 5
OP_SET_PERMISSIONS ( 7): 4
OP_ADD ( 0): 2
OP_RENAME_OLD ( 1): 2
OP_CLOSE ( 9): 2
OP_SET_GENSTAMP_V2 ( 31): 2
OP_ALLOCATE_BLOCK_ID ( 32): 2
OP_ADD_BLOCK ( 33): 2
OP_SET_OWNER ( 8): 1
OP_END_LOG_SEGMENT ( 23): 1
OP_START_LOG_SEGMENT ( 24): 1