Metrics

Alluxio Metrics

Cache Storage

Metric	Labels	Type	Component	Description
alluxio_cached_storage	type	gauge	worker	amount of the cached data
alluxio_cached_capacity	type	gauge	worker	configured maximum cache storage
alluxio_data_cached_pages	-	gauge	worker	number of pages being cached in the page store
alluxio_data_cached_files	-	gauge	worker	number of cached files, including fully and partially cached files
alluxio_cached_storage_by_priority	priority	gauge	worker	amount of the cached data
alluxio_eviction_by_ttl	policy_path	counter	worker	Total number of bytes evicted from Alluxio workers by TTL policy.
alluxio_quota_size_used	dir	gauge	coordinator	Bytes used in the given quota scope.
alluxio_quota_size_capacity	dir	gauge	coordinator	Capacity of the given quota scope as defined in the quota rules.
alluxio_eviction_by_quota	dir	counter	worker	Total number of bytes evicted from Alluxio workers by quota rules.

Cache Access

Metric	Labels	Type	Component	Description
alluxio_data_access	method	histogram	worker	aggregated all data access requests
alluxio_data_throughput	dir, method, destination	counter	worker	counter of data throughput of all data access
alluxio_meta_operation	op	counter	worker	counter of rpc calls of the meta operations
alluxio_meta_operation_latency_ms	op, state	histogram	worker	latency of rpc calls of the meta operations
alluxio_meta_operation_errors	op	counter	worker	counter of errors during handling of rpc calls of the meta operations
alluxio_ufs_error	ufs_type, error_code	counter	worker	counter of the rpc calls
alluxio_ufs_latency_ms	method, ufs_type	histogram	worker	Histogram of ufs call latency
alluxio_ufs_client_latency_ms	method, ufs_type, state	histogram	worker	Histogram of ufs client api call latency
alluxio_ufs_client_call_processing	method, ufs_type	gauge	worker	Gauge of the ufs client calls that are being processed
alluxio_ufs_data_access	dir, method	counter	worker	amount of the ufs access
alluxio_ufs_fallback	method	counter	worker	amount of the ufs fallback
alluxio_cached_data_read	dir, is_pinned	counter	worker	amount of data that, when read, was present in and served from the Alluxio cache
alluxio_missed_data_read	dir, is_pinned	counter	worker	amount of data that, when read, was absent from the Alluxio cache
alluxio_cache_hit_calls	-	counter	worker	number of cache hits in page store
alluxio_cache_miss_calls	-	counter	worker	number of cache misses in page store
alluxio_external_data_read	dir	counter		amount of the read data when cache missed on client
alluxio_cleared_stale_cached_data	-	counter	worker	amount of cleared stale cached data
alluxio_cached_evicted_data	-	counter	worker	amount of the evicted data
alluxio_cached_async_evicted_data	-	counter	worker	amount of the async evicted data
alluxio_trigger_async_evicted_total	-	counter	worker	counter of times asynchronous eviction is triggered
alluxio_page_store_operation_errors	op, cause	counter	worker	counter of failures in page store operations
alluxio_page_store_dir_operation_errors	dir	counter	worker	counter of failures in specific page store directory
alluxio_page_store_dir_operations	dir	counter	worker	operation counter in specific page store directory
alluxio_page_store_io_latency_microseconds	dir, op, success	histogram	worker	latency of IO operations in page store
alluxio_metadata_cache_hit_calls	type	counter	worker	counter of metadata retrieval calls that resulted in a cache hit
alluxio_external_file_metadata_request_calls	-	counter	worker	counter of file metadata retrieval calls that are fetched from UFS, usually as a result of a cache miss in the file metadata cache
alluxio_metadata_cache_miss_calls	type	counter	worker	counter of metadata retrieval calls that resulted in a cache miss. Note that this is different from alluxio_external_file_metadata_request_calls in that a cache miss for a file does not always result in a request to an external data source.
alluxio_passive_cache_async_loaded_files	result	counter	worker	number of async loaded files when passive cache is enabled
alluxio_page_store_device_total_capacity	dir	gauge	worker	the total capacity of the physical storage device where the page store directory resides
alluxio_page_store_device_available_capacity	dir	gauge	worker	the available capacity of the physical storage device where the page store directory resides
alluxio_metastore_storage_size	dir	gauge	worker	Total logical size of files in the metastore RocksDB directory.
alluxio_metastore_disk_capacity	dir	gauge	worker	The capacity of the disk where the metastore rocksdb is located.
alluxio_netty_data_ingress	-	counter	worker	number of ingress bytes from clients to worker, excluding TLS
alluxio_netty_data_egress	-	counter	worker	number of egress bytes from worker to clients, excluding TLS
alluxio_worker_thread_pool_rejections	dir	counter	worker	counter of rejections in worker thread pool
alluxio_rpc_executor_current_queue_length	executor_name	gauge	worker	number of RPC requests currently being processed and pending processing
alluxio_rpc_executor_active_threads	executor_name	gauge	worker	number of threads that are actively executing RPCs
alluxio_rpc_executor_current_threads	executor_name	gauge	worker	number of threads for executing RPCs, both occupied and idle
alluxio_rpc_executor_max_threads	executor_name	gauge	worker	maximum number of threads for executing RPCs

S3 API

Metric	Labels	Type	Component	Description
alluxio_s3_api_throughput	method	histogram	worker	histogram of S3 API throughput
alluxio_s3_api_call_latency_ms	method, state	histogram	worker	latency of S3 API calls
alluxio_s3_api_call_processing	method	gauge	worker	counter of the S3 API calls that are being processed
alluxio_s3_authn_latency_ms	result, reason	histogram	worker	Latency of S3 authentication.
alluxio_s3_authz_latency_ms	result, method, reason	histogram	worker	Latency of S3 authorization.
alluxio_sts_api_call_processing	method	gauge	worker	Counter of the sts API calls that are being processed.
alluxio_sts_requests_total	result, reason	counter	worker	Total number of STS requests.

FUSE

Metric	Labels	Type	Component	Description
alluxio_fuse_concurrency	method	gauge	fuse	record the realtime concurrency for fuse method
alluxio_fuse_call_latency_ms	method, state	histogram	fuse	latency of fuse operations
alluxio_fuse_result	method, state	counter	fuse	counter of fuse operation results
alluxio_fuse_path_cache_hits_bytes	-	counter	fuse	counter of fuse path cache hits
alluxio_fuse_path_cache_misses	-	counter	fuse	counter of fuse path cache misses
alluxio_fuse_buffer_size	method, sequential	histogram	fuse	Record sequential or random read/write and its buffer size.
alluxio_fuse_block_size	method	histogram	fuse	Record the block size during random reads/writes. 'Block size' can be understood as the 'bs' parameter specified during fio testing.
alluxio_fuse_open_files	-	gauge	fuse	The number of fuse open files.

Client SDK

Metric	Labels	Type	Component	Description
alluxio_grpc_client_call_latency_ms	method, instance, state	histogram	worker, coordinator, fuse	latency of gRPC calls from the client
alluxio_grpc_client_concurrency	method, instance	gauge	worker, coordinator, fuse	concurrency of gRPC calls from the client
alluxio_grpc_client_errors	method, status_code, instance	counter	worker, coordinator, fuse	total number of gRPC errors from the client
alluxio_grpc_client_successes	method, instance	counter	worker, coordinator, fuse	total number of successful gRPC calls from the client
alluxio_netty_operations	op	counter	worker, coordinator, fuse	number of netty operations (e.g. read and write requests)
alluxio_netty_operation_errors	op, reason, instance	counter	worker, coordinator, fuse	total number of Netty operation errors from the client
alluxio_read_from_workers	instance	counter	worker, coordinator, fuse	total number of client read bytes from worker
alluxio_async_prefetch_cache_bytes	instance	counter	worker, coordinator, fuse	total number of bytes that client async prefetch data to local
alluxio_async_prefetch_hit_cache_bytes	instance	counter	worker, coordinator, fuse	total number of bytes that client hit cache from async prefetch cache
alluxio_async_prefetch_random_read_requests	instance	counter	worker, coordinator, fuse	total number of client random read recorded by async prefetch
alluxio_multi_replica_read_from_workers	cluster_name, local_cluster, hot_read	counter	worker	number of bytes read by a client from Alluxio workers when reading multi-replica files
alluxio_rpc_retry_on_different_workers	op, retry_count	counter	worker	counter of client retry on different workers if multi replica is enabled.
alluxio_rpc_position_reader_read_calls	component	counter	worker	counter of client position reader read success.
alluxio_rpc_position_reader_data_read	component	counter	worker	counter of bytes read by client position reader.
alluxio_rpc_position_reader_read_failed_total	component, final_attempt	counter	worker	counter of client position reader read failure.
alluxio_client_netty_read_time_to_receive_first_packet_ms	-	histogram	fuse	latency between when the client sends a read request to the worker, and when the worker sends the first packet of the response back to the client.

Job Service

Metric	Labels	Type	Component	Description
alluxio_completed_job	type, state	counter	coordinator	counter of the jobs.
alluxio_job_process_file	type, state	counter	coordinator	counter of the files.
alluxio_job_process_file_size	type, state	counter	coordinator	cumulative size of the files that are processed by job service.
alluxio_active_job_count	type	gauge	coordinator	counter of the jobs in scheduler. the value of type is running or waiting
alluxio_distributed_load_job_dispatched_size	-	counter	coordinator	counter of the bytes dispatched in distributed load
alluxio_distributed_load_job_failure	reason, final_attempt, worker	counter	coordinator	counter of the distributed load failure
alluxio_distributed_load_job_loaded_bytes	-	counter	coordinator	counter of the bytes loaded in distributed load
alluxio_distributed_load_job_processed	-	counter	coordinator	counter of the non empty file copies loaded in distributed load
alluxio_distributed_load_job_scanned	-	counter	coordinator	counter of the inodes scanned in distributed load
alluxio_distributed_load_job_skipped	-	counter	coordinator	counter of the inodes skipped in distributed load
alluxio_distributed_load_data_loaded	-	counter	worker	counter of the bytes loaded by a worker in distributed load
alluxio_distributed_load_data_loaded_from_ufs	-	counter	worker	counter of the bytes loaded by a worker from ufs in distributed load
alluxio_worker_job_task_count	-	gauge	coordinator	Number of tasks currently executed by each worker.

Write Cache

Metric	Labels	Type	Component	Description
alluxio_write_buffer_write_status	status	counter	worker	the status of write buffer writes
alluxio_write_buffer_worker_failure	worker	counter	worker	the failure count of writes to workers
alluxio_write_buffer_worker_bytes_written	worker	counter	worker	the bytes written to workers
alluxio_write_buffer_unique_bytes_written	-	counter	worker	the unique bytes written by the client
alluxio_write_buffer_foundationdb_call_latency_ms	method, state	histogram	worker	the latency of FoundationDB calls
alluxio_write_buffer_persist_tasks	status	counter	worker	the number of persist tasks
alluxio_write_buffer_transition_worker	worker	counter	worker	the number of worker transitions
alluxio_write_buffer_async_persist_throughput	-	counter	worker	the throughput of async persist
alluxio_write_buffer_async_file_checker_abnormal_files	-	counter	worker	the number of abnormal files found by the async file checker
alluxio_dual_buffer_file_system_requests	operation, buffer_type	counter	worker	counter of requests to dual buffer file system.

Authorization

Metric	Labels	Type	Component	Description
alluxio_auth_permission_check_total	-	counter	worker, fuse	The total number of authorization permission checks.
alluxio_auth_permission_check_cache_misses	-	counter	worker, fuse	The total number of misses in the authorization permission check cache.

Cluster & Process

Metric	Labels	Type	Component	Description
alluxio_version	version	gauge	worker, coordinator, fuse	Alluxio component version information
alluxio_license_expiration_date	-	gauge	coordinator	the license expiration date in epoch time format
alluxio_cumulative_unavailable_workers	worker_addr	counter	worker, coordinator	number of cumulative occurrences of unavailable workers encountered by a client.
alluxio_unavailable_worker_probe_attempts	worker_addr, attempt	counter	worker, coordinator	number of liveness probing attempts for an unavailable workers
alluxio_worker_membership_refresh_count	-	counter	worker	total number of worker membership refreshes
alluxio_dynamic_resource_pool_current_resources	pool_name	gauge	worker	current number of resources in the dynamic resource pool
alluxio_dynamic_resource_pool_capacity	pool_name	gauge	worker	capacity of the dynamic resource pool
alluxio_dynamic_resource_pool_acquisition_timeouts	pool_name	counter	worker	total number of acquisition timeouts in the dynamic resource pool
alluxio_dynamic_resource_pool_create_new_resource_latency_ms	pool_name	histogram	worker	latency of creating a new resource in the dynamic resource pool
alluxio_etcd_call_errors	type	counter	coordinator, worker, fuse	total number of etcd call errors
alluxio_etcd_client_calls	type	counter	coordinator, worker, fuse	total number of etcd client calls
alluxio_etcd_client_call_latency_ms	type	histogram	coordinator, worker, fuse	latency of etcd client calls
alluxio_netty_direct_memory_usage	-	gauge	worker	direct memory usage of Netty
alluxio_rocksdb_memory_usage	-	gauge	worker	memory usage of RocksDB
process_start_time_seconds	-	gauge	coordinator, worker, fuse	start time of the process since unix epoch in seconds
process_cpu_seconds_total	-	counter	coordinator, worker, fuse	total user and system CPU time spent in seconds
jvm_threads_current	-	gauge	coordinator, worker, fuse	current thread count of a JVM
jvm_memory_used_bytes	area=heap/nonheap	gauge	coordinator, worker, fuse	used bytes of a given JVM memory area
jvm_memory_max_bytes	area=heap/nonheap	gauge	coordinator, worker, fuse	max (bytes) of a given JVM memory area
jvm_gc_collection_seconds	gc="G1 Young Generation"/"G1 Old Generation"/...	summary	coordinator, worker, fuse	time spent in a given JVM garbage collector in seconds
jvm_buffer_pool_used_bytes	pool=direct/mapped	gauge	coordinator, worker, fuse	used bytes of a given JVM buffer pool