org.hbase.async

Class ClientStats

java.lang.Object

org.hbase.async.ClientStats

public final class ClientStats
extends Object

HBaseClient usage statistics.

This is an immutable snapshot of usage statistics of the client. Please note that not all the numbers in the snapshot are collected atomically, so although each individual number is up-to-date as of the time this object is created, small inconsistencies between numbers can arise.

Since:

1.3

Method Summary

Methods

Modifier and Type	Method and Description
long	appends() Number calls to HBaseClient.append(org.hbase.async.AppendRequest).
long	atomicIncrements() Number of AtomicIncrementRequest sent.
long	connectionsCreated() Number of connections created to connect to RegionServers.
long	contendedMetaLookups() Returns how many lookups in .META. were performed (contended).
int	deadRegionClients() The number of region clients that have lost their connection to the region server.
long	deletes() Number calls to HBaseClient.delete(org.hbase.async.DeleteRequest).
long	flushes() Number of calls to HBaseClient.flush().
long	gets() Number of calls to HBaseClient.get(org.hbase.async.GetRequest).
long	idleConnectionsClosed() Returns the number of connections to region servers that were closed due to being idle past the "hbase.hbase.ipc.client.connection.idle_timeout" value.
com.google.common.cache.CacheStats	incrementBufferStats() Returns statistics from the buffer used to coalesce increments.
long	inflightRPCs() Represents the number of RPCs that have been sent to the region client and are currently waiting for a response.
long	noSuchRegionExceptions() Number of NoSuchRegionException handled by the client.
long	numBatchedRpcSent() Number of batched RPCs sent to the network.
long	numRpcDelayedDueToNSRE() Number of RPCs delayed due to NoSuchRegionExceptions.
long	pendingBatchedRPCs() The number of batched RPCs waiting to be sent to the server.
long	pendingRPCs() The number of RPCs that are queued up and ready to be sent to the region server.
long	puts() Number calls to HBaseClient.put(org.hbase.async.PutRequest).
int	regionClients() The number of instantiated region clients.
long	rootLookups() Returns how many lookups in -ROOT- were performed.
long	rowLocks() Number calls to HBaseClient.lockRow(org.hbase.async.RowLockRequest).
long	scannersOpened() Number of scanners opened.
long	scans() Number of times a scanner had to fetch data from HBase.
long	uncontendedMetaLookups() Returns how many lookups in .META. were performed (uncontended).

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

connectionsCreated

public long connectionsCreated()

Number of connections created to connect to RegionServers.

idleConnectionsClosed

public long idleConnectionsClosed()

Returns the number of connections to region servers that were closed due to being idle past the "hbase.hbase.ipc.client.connection.idle_timeout" value.

Returns:

The number of idle connections over time

Since:

1.7

rootLookups

public long rootLookups()

Returns how many lookups in -ROOT- were performed.

This number should remain low. It will be 1 after the first access to HBase, and will increase by 1 each time the .META. region moves to another server, which should seldom happen.

This isn't to be confused with the number of times we looked up where the -ROOT- region itself is located. This happens even more rarely and a message is logged at the INFO whenever it does.

uncontendedMetaLookups

public long uncontendedMetaLookups()

Returns how many lookups in .META. were performed (uncontended).

This number indicates how many times we had to lookup in .META. where a key was located. This only counts "uncontended" lookups, where the thread was able to acquire a "permit" to do a .META. lookup. The majority of the .META. lookups should fall in this category.

contendedMetaLookups

public long contendedMetaLookups()

Returns how many lookups in .META. were performed (contended).

This number indicates how many times we had to lookup in .META. where a key was located. This only counts "contended" lookups, where the thread was unable to acquire a "permit" to do a .META. lookup, because there were already too many .META. lookups in flight. In this case, the thread was delayed a bit in order to apply a bit of back-pressure on the caller, to avoid creating .META. storms. The minority of the .META. lookups should fall in this category.

flushes

public long flushes()

Number of calls to HBaseClient.flush().

noSuchRegionExceptions

public long noSuchRegionExceptions()

Number of NoSuchRegionException handled by the client.

The NoSuchRegionException is an integral part of the way HBase work. HBase clients keep a local cache of where they think each region is in the cluster, but in practice things aren't static, and regions will move due to load balancing, or get split into two new regions due to write activity. When this happens, clients find out "the hard way" that their RPC failed because the region they tried to get to is no longer there. This causes the client to invalidate its local cache entry for this region and perform a .META. lookup to find where this region has moved, or find the new region to use in case of a split.

While NoSuchRegionException are expected to happen due to load balancing or write load, they tend to have a large performance impact as they force the clients to back off and repeatedly poll the cluster to find the new location of the region. So it's good to keep an eye on the rate at which they happen to make sure it remains fairly constant and low.

In a high write throughput application, if this value increases too quickly it typically indicates that there are too few regions, so splits are happening too often. In this case you should manually split the hot regions in order to better distribute the write load.

This number is a subset of numRpcDelayedDueToNSRE(), because each NoSuchRegionException causes multiple RPCs to be delayed.

numRpcDelayedDueToNSRE

public long numRpcDelayedDueToNSRE()

Number of RPCs delayed due to NoSuchRegionExceptions.

In a high throughput application, if this value increases too quickly it typically indicates that there are too few regions, or that some regions are too hot, which is causing too many RPCs to back up when a region becomes temporarily unavailable. In this case you should manually split the hot regions in order to better distribute the write load.

See Also:

noSuchRegionExceptions()

numBatchedRpcSent

public long numBatchedRpcSent()

Number of batched RPCs sent to the network.

While puts() and deletes() indicate the number of RPCs created at the application level, they don't reflect the actual number of RPCs sent to the network because of batching (see HBaseClient.setFlushInterval(short)).

Note that deletes() can only be batched if you use HBase 0.92 or above.

gets

public long gets()

Number of calls to HBaseClient.get(org.hbase.async.GetRequest).

scannersOpened

public long scannersOpened()

Number of scanners opened.

scans

public long scans()

Number of times a scanner had to fetch data from HBase.

puts

public long puts()

Number calls to HBaseClient.put(org.hbase.async.PutRequest).

Note that this doesn't necessarily reflect the number of RPCs sent to HBase due to batching (see HBaseClient.setFlushInterval(short)).

See Also:

numBatchedRpcSent()

appends

public long appends()

Number calls to HBaseClient.append(org.hbase.async.AppendRequest).

Note that this doesn't necessarily reflect the number of RPCs sent to HBase due to batching (see HBaseClient.setFlushInterval(short)).

See Also:

numBatchedRpcSent()

rowLocks

public long rowLocks()

Number calls to HBaseClient.lockRow(org.hbase.async.RowLockRequest).

deletes

public long deletes()

Number calls to HBaseClient.delete(org.hbase.async.DeleteRequest).

Note that if you use HBase 0.92 or above, this doesn't necessarily reflect the number of RPCs sent to HBase due to batching (see HBaseClient.setFlushInterval(short)).

See Also:

numBatchedRpcSent()

atomicIncrements

public long atomicIncrements()

Number of AtomicIncrementRequest sent.

This number includes AtomicIncrementRequests sent after being buffered by HBaseClient.bufferAtomicIncrement(org.hbase.async.AtomicIncrementRequest). The number of evictions returned by incrementBufferStats() is a subset of this number, and the difference between the two is the number of increments that were sent directly without being buffered.

inflightRPCs

public long inflightRPCs()

Represents the number of RPCs that have been sent to the region client and are currently waiting for a response. If this value increases then the region server is likely overloaded.

Returns:

the number of RPCs sent to region client waiting for response.

Since:

1.7

pendingRPCs

public long pendingRPCs()

The number of RPCs that are queued up and ready to be sent to the region server. When an RPC is sent, this number should be decremented and inflightRPCs incremented.

Returns:

the number of RPCs queued and ready to be sent to region server.

Since:

1.7

pendingBatchedRPCs

public long pendingBatchedRPCs()

The number of batched RPCs waiting to be sent to the server.

Returns:

the number of batched RPCs waiting to be sent to server.

Since:

1.7

deadRegionClients

public int deadRegionClients()

The number of region clients that have lost their connection to the region server.

Returns:

the number of region clients that have lost connection to region server.

Since:

1.7

regionClients

public int regionClients()

The number of instantiated region clients.

Returns:

the number of instantiated region clients.

Since:

1.7

incrementBufferStats

public com.google.common.cache.CacheStats incrementBufferStats()

Returns statistics from the buffer used to coalesce increments.