oracle rac instacne processes vs single-instance

zr_baobao

Oracle 11g rac administrator

 

this chapter focuses on administration of your database when it’s running in an Oracle RAC environment,you will learn the differences between a single-instance database and an Oracle RAC database,how to tune effectively in an Oracle RAC environment,and how to patch an Oracle RAC database.in addition,the chapter covers details of using the SRVCTL utility to manger your Oracle RAC environment,and wraps up with a dicussion of serer policies and the manging of diagnostic data.

 

Oracle RAC vs Single –instance: addition processes RAC数据库和单实例数据的进程之前的区别

 

in an Oracle RAC environment,there are several additional background process associated with each Oracle RAC instance,beyond what whould normally be seen in a single-instance environment.these processes work together in a coordinated fashion to maintain the locks necessary for multiple instaces to access resources simultaneously ,and to ensure that these resource are made available to the instances where they are most needed,in a timely fashion.

a simple query such as the following will give you an idea of the background processes involved in an Oracle RAC environment:

set pages 50

select name,descriptin from v$bgprocess where PADDR<>’00’;

 

Comparing the list of processes between Oracle RAC and single-instance environments,you will see that there are several which are specific to an Oracle RAC instance,A brief description of these unique processes and how they interact an an Oracle RAC environment is provided next.

 

LCK:Lock Process

the lock process(LCK) manages requests that are not cache-fusion requests,such as row cache and library cache requests,Only a single LCK process is allowed for each instacne.LCK mantains a list of lock elements and uses this list to validate locks during instacne recovery.

 

LMD: Lock Manager Daemon Process //锁管理监控进程

The lock manager daemon(LMD) process is also known as the global enqueue service daemon (全局enqueue服务监控 GESD).From within each instance ,the LMD process manages incoming remote resource request(LMD管理远程资源的请求).It is also responsible for deadlock detection and monitoring for lock conversion timeouts.(同样会对死锁的发现以及锁转换超时的探测).

 

LMON:lock monitor process

the lock monitor (LMON) process is the global enqueue service monitor(GESM),it is responsible for the reconfiguration of lock resources when an instacne joins the cluster or leaves the cluster and is also responsible for the dynamic lock remastering mentioned earlier.LMON will generate a trace file whenever a reconfiguration occurs(as opposed to remastering of a subset of locks).it is the responsibility of LMON to check for the death of instances clusterwide and to initiate reconfiguration as quicklu as possible.

 

LMS:Lock Manager Server Process

the lock manager server (LMS,or global cache service process GCS) process is in charge of shipping the blocks between instances for cache-fusion requests. in the event of a consistent-read request,the LMS process will first rollback the blocks,creating the consistent read(CR)  image of the block ,and will then ship that version of the block across the interconnect to the foreground process making the request at the remote instance,in addition,LMS must interact with the LMD process to retrieve lock request placed by LMD.and instacne may dynamically generate up to ten LMS processes;the number of LMS processes can be set with the parameter GCS_SERVER_PROCESSES,and the value is dependent on the number of CPUS.at startup time,CPU_COUNT/4 LMS processes are started ,but at least two LMS processes are always started .from Oracle Database version 10.2 and on,only one LMS process is started if the server just has one CPU.if you are consolidating many small databases into a cluster,you may want to reduce the number of Global Cache Service(GCS) processes (LMSn) created by the Oracle RAC instance.to ensure the LMS processes get CPU time when needed,the LMS processes must run in a scheduling priority set to Real Time.

 

ACFS :ASM Cluster File System CSS process

the automatic storage management cluster file system CSS process.delivers CSS membership changes to the ASM cluster file system.these membership changes are required for the file system to maintain file system consistency within the cluster,this process is new with Oracle Database 11g Release 2.

 

ACMS:Atimic Control Fiel to Memory Service Process

the atomic Control File to memory Service(ACMS) process works with a coordinating caller to ensure that an operation is executed on every instance in Oracle RAC despite failures.ACMS is the process in which a distributed operation is called.as a result,this process can exhibit a variety of beahviors,in general,ACMS is limited to small,nonblocking state changes for a limited set of cross-instance operations,this process is new with 11g release2.

 

 

 

GTXn:Global transaction Process

Global Transaction (GTXn) processes help maintain the global informatioin about (XA) global transactions throughout the cluster,also,the processes help perform two-phase commit for global transactions anywhere in the cluster so that an Oracle RAC database behaves as a single system to the externally coordinated distributed transactions.this process is new with Oracle Database 11g release2.

 

 

LMHB:global cache /enqueue service hearbeat monitor

 

another process new with 11grelease 2,LM heartbeat monitor(LMHB)monitors LMON,LMD ,and LMSn processes to ensures that they are running normally without blocking or spinging.

 

PING:interconnect latency measurement process

every few seconds ,the PING process in one instacne sends messages to each instance.the message is received by PING on the target instance.the time for the round trip is measured and collected,this is also a new process with 11g release 2.

 

RMSn:Oracle RAC Management Process

the oracle rac management(RMSn) Process performs a variety of tasks,including creating resources related to Oracle RAC when new instances are added to a cluster.

RSMN:Remote Slave Monitor Process

the remote Slave monitor(RSMN) background process manages the creation of slave processes and the communication with their coordinators and peers.these background slave processes perform tasks an behalf of acoordinating process running in another cluster instance.