Aerospike Monitoring

Aerospike is a specific type of distributed, scalable database that was built with three core objectives in mind. First, its developers wanted to create a flexible, scalable platform that was suitable to meet the demands of modern web-scale applications. Next, they aimed to provide the robustness and reliability expected from traditional databases, all while taking things one step further. Finally, they wanted to provide operational efficiency to organizations of all types, all with the bare minimum of manual involvement required.

In a larger sense, Aerospike is a client-server solution that allows users to build an app server cluster that itself attaches over TCP to your database cluster. All of this is possible using Aerospike's built-in libraries. Keep in mind that for high scale Internet applications, for example, a key-value database depends on persistence, sharding, object operations (like list and map manipulation) and other important qualities. All of this is in service of empowering the great operational characteristics that users depend on. To get to that point, a wide range of application languages need to be supported as most teams use not just a single one, but a combination of many.

Aerospike is suitable for all of these applications and more, all while acting as one of the fastest key-value databases available today. Note that Aerospike is also highly scalable by design, and it's very easy to get up and running because all nodes are the same without any master/slave setup or other differences to try to figure out on your end.

Aerospike database technology is centered around row-based random access with indexes in memory and data in memory or on SSD (solid-state drive) storage. The database holds data that is accessible in real time.

Why use a Telegraf plugin for Aerospike?

The Telegraf input plugin for Aerospike allows users to easily query Aerospike servers in real-time, all in the name of gathering important information like node statistics and statistics for all configured namespaces. Then, all of that information is condensed into a series of straightforward dashboards, visualizing things in a way that reveals important information and other insights that likely would have otherwise gone undiscovered.

The major benefit is that not only is this information easy-to-understand, but it's also even easier to share - making sure that everyone within your organization has access to all the important data they need to do their jobs at all times.

How to monitor Aerospike using the Telegraf plugin

Configuring the Aerospike Telegraf Plugin is simple. Add the servers, credentials and the optional TLS configurations, and you will be ingesting your Aerospike metrics and events into InfluxDB in no time. The metrics collected are under two (InfluxDB) measurements:

1. aerospike_node: These are the Aerospike node measurements, which are available from the Aerospike statistics command.
2. aerospike_namespace: These are Aerospike namespace measurements, which are available from the Aerospike namespace/<namespace_name> command.

All measurements have tags of aerospike_host and node_name, and the Namespace metrics have tags of namespace_name.

Example output

% telegraf --input-filter aerospike --test
> aerospike_node,aerospike_host=localhost:3000,node_name="BB9020011AC4202" batch_error=0i,batch_index_complete=0i,batch_index_created_buffers=0i,batch_index_destroyed_buffers=0i,batch_index_error=0i,batch_index_huge_buffers=0i,batch_index_initiate=0i,batch_index_queue="0:0,0:0,0:0,0:0",batch_index_timeout=0i,batch_index_unused_buffers=0i,batch_initiate=0i,batch_queue=0i,batch_timeout=0i,client_connections=6i,cluster_integrity=true,cluster_key="8AF422E05281249E",cluster_size=1i,delete_queue=0i,demarshal_error=0i,early_tsvc_batch_sub_error=0i,early_tsvc_client_error=0i,early_tsvc_udf_sub_error=0i,fabric_connections=16i,fabric_msgs_rcvd=0i,fabric_msgs_sent=0i,heartbeat_connections=0i,heartbeat_received_foreign=0i,heartbeat_received_self=0i,info_complete=47i,info_queue=0i,migrate_allowed=true,migrate_partitions_remaining=0i,migrate_progress_recv=0i,migrate_progress_send=0i,objects=0i,paxos_principal="BB9020011AC4202",proxy_in_progress=0i,proxy_retry=0i,query_long_running=0i,query_short_running=0i,reaped_fds=0i,record_refs=0i,rw_in_progress=0i,scans_active=0i,sindex_gc_activity_dur=0i,sindex_gc_garbage_cleaned=0i,sindex_gc_garbage_found=0i,sindex_gc_inactivity_dur=0i,sindex_gc_list_creation_time=0i,sindex_gc_list_deletion_time=0i,sindex_gc_locktimedout=0i,sindex_gc_objects_validated=0i,sindex_ucgarbage_found=0i,sub_objects=0i,system_free_mem_pct=92i,system_swapping=false,tsvc_queue=0i,uptime=1457i 1468923222000000000
> aerospike_namespace,aerospike_host=localhost:3000,namespace=test,node_name="BB9020011AC4202" allow_nonxdr_writes=true,allow_xdr_writes=true,available_bin_names=32768i,batch_sub_proxy_complete=0i,batch_sub_proxy_error=0i,batch_sub_proxy_timeout=0i,batch_sub_read_error=0i,batch_sub_read_not_found=0i,batch_sub_read_success=0i,batch_sub_read_timeout=0i,batch_sub_tsvc_error=0i,batch_sub_tsvc_timeout=0i,client_delete_error=0i,client_delete_not_found=0i,client_delete_success=0i,client_delete_timeout=0i,client_lang_delete_success=0i,client_lang_error=0i,client_lang_read_success=0i,client_lang_write_success=0i,client_proxy_complete=0i,client_proxy_error=0i,client_proxy_timeout=0i,client_read_error=0i,client_read_not_found=0i,client_read_success=0i,client_read_timeout=0i,client_tsvc_error=0i,client_tsvc_timeout=0i,client_udf_complete=0i,client_udf_error=0i,client_udf_timeout=0i,client_write_error=0i,client_write_success=0i,client_write_timeout=0i,cold_start_evict_ttl=4294967295i,conflict_resolution_policy="generation",current_time=206619222i,data_in_index=false,default_ttl=432000i,device_available_pct=99i,device_free_pct=100i,device_total_bytes=4294967296i,device_used_bytes=0i,disallow_null_setname=false,enable_benchmarks_batch_sub=false,enable_benchmarks_read=false,enable_benchmarks_storage=false,enable_benchmarks_udf=false,enable_benchmarks_udf_sub=false,enable_benchmarks_write=false,enable_hist_proxy=false,enable_xdr=false,evict_hist_buckets=10000i,evict_tenths_pct=5i,evict_ttl=0i,evicted_objects=0i,expired_objects=0i,fail_generation=0i,fail_key_busy=0i,fail_record_too_big=0i,fail_xdr_forbidden=0i,geo2dsphere_within.earth_radius_meters=6371000i,geo2dsphere_within.level_mod=1i,geo2dsphere_within.max_cells=12i,geo2dsphere_within.max_level=30i,geo2dsphere_within.min_level=1i,geo2dsphere_within.strict=true,geo_region_query_cells=0i,geo_region_query_falsepos=0i,geo_region_query_points=0i,geo_region_query_reqs=0i,high_water_disk_pct=50i,high_water_memory_pct=60i,hwm_breached=false,ldt_enabled=false,ldt_gc_rate=0i,ldt_page_size=8192i,master_objects=0i,master_sub_objects=0i,max_ttl=315360000i,max_void_time=0i,memory_free_pct=100i,memory_size=1073741824i,memory_used_bytes=0i,memory_used_data_bytes=0i,memory_used_index_bytes=0i,memory_used_sindex_bytes=0i,migrate_order=5i,migrate_record_receives=0i,migrate_record_retransmits=0i,migrate_records_skipped=0i,migrate_records_transmitted=0i,migrate_rx_instances=0i,migrate_rx_partitions_active=0i,migrate_rx_partitions_initial=0i,migrate_rx_partitions_remaining=0i,migrate_sleep=1i,migrate_tx_instances=0i,migrate_tx_partitions_active=0i,migrate_tx_partitions_imbalance=0i,migrate_tx_partitions_initial=0i,migrate_tx_partitions_remaining=0i,non_expirable_objects=0i,ns_forward_xdr_writes=false,nsup_cycle_duration=0i,nsup_cycle_sleep_pct=0i,objects=0i,prole_objects=0i,prole_sub_objects=0i,query_agg=0i,query_agg_abort=0i,query_agg_avg_rec_count=0i,query_agg_error=0i,query_agg_success=0i,query_fail=0i,query_long_queue_full=0i,query_long_reqs=0i,query_lookup_abort=0i,query_lookup_avg_rec_count=0i,query_lookup_error=0i,query_lookup_success=0i,query_lookups=0i,query_reqs=0i,query_short_queue_full=0i,query_short_reqs=0i,query_udf_bg_failure=0i,query_udf_bg_success=0i,read_consistency_level_override="off",repl_factor=1i,scan_aggr_abort=0i,scan_aggr_complete=0i,scan_aggr_error=0i,scan_basic_abort=0i,scan_basic_complete=0i,scan_basic_error=0i,scan_udf_bg_abort=0i,scan_udf_bg_complete=0i,scan_udf_bg_error=0i,set_deleted_objects=0i,sets_enable_xdr=true,sindex.data_max_memory="ULONG_MAX",sindex.num_partitions=32i,single_bin=false,stop_writes=false,stop_writes_pct=90i,storage_engine="device",storage_engine.cold_start_empty=false,storage_engine.data_in_memory=true,storage_engine.defrag_lwm_pct=50i,storage_engine.defrag_queue_min=0i,storage_engine.defrag_sleep=1000i,storage_engine.defrag_startup_minimum=10i,storage_engine.disable_odirect=false,storage_engine.enable_osync=false,storage_engine.file="/opt/aerospike/data/test.dat",storage_engine.filesize=4294967296i,storage_engine.flush_max_ms=1000i,storage_engine.fsync_max_sec=0i,storage_engine.max_write_cache=67108864i,storage_engine.min_avail_pct=5i,storage_engine.post_write_queue=0i,storage_engine.scheduler_mode="null",storage_engine.write_block_size=1048576i,storage_engine.write_threads=1i,sub_objects=0i,udf_sub_lang_delete_success=0i,udf_sub_lang_error=0i,udf_sub_lang_read_success=0i,udf_sub_lang_write_success=0i,udf_sub_tsvc_error=0i,udf_sub_tsvc_timeout=0i,udf_sub_udf_complete=0i,udf_sub_udf_error=0i,udf_sub_udf_timeout=0i,write_commit_level_override="off",xdr_write_error=0i,xdr_write_success=0i,xdr_write_timeout=0i,{test}_query_hist_track_back=300i,{test}_query_hist_track_slice=10i,{test}_query_hist_track_thresholds="1,8,64",{test}_read_hist_track_back=300i,{test}_read_hist_track_slice=10i,{test}_read_hist_track_thresholds="1,8,64",{test}_udf_hist_track_back=300i,{test}_udf_hist_track_slice=10i,{test}_udf_hist_track_thresholds="1,8,64",{test}_write_hist_track_back=300i,{test}_write_hist_track_slice=10i,{test}_write_hist_track_thresholds="1,8,64" 1468923222000000000

Key Aerospike metrics to use for monitoring

Aerospike recommends monitoring a set of key metrics that are divided into 6 categories. Each category indicates a common system component that may cause the metric to report an alert/critical value. Some of the important Aerospike metrics that you should proactively monitor include:

• Application metrics that may indicate issues in the application layer
• Memory metrics that may be used to indicate abnormal memory utilization
• Network metrics that may indicate problems on the network layer
• Storage metrics that may be used to indicate abnormal disk utilization
• Service/other metrics that indicate abnormal database operation or system problems that may cause abnormal database operations
• Trend metrics, which are useful stats to allow for deeper understanding of system behaviors leading up to a particular event

Project URL   Documentation