HACMP
HACMP stands for High Availability Cluster Multiprocessing
Hacmp provides two types of environment
1. serial (high availability): this is used to make the application highlt available
by using the shared disk and duplicate resources . also the access to data will be serially
means one node at a time.
2. parallel (cluster multiprocessing ) : this type of environment is used when the application is
online on many nodes and allthe nodes can concurrently access the data. for this type of environment
failover is not required.
why to use hacmp?
1. eliminating single point of failure
2. elminating planned and unplanned downtime.
hacmp ensures that the application should be highly available and accessible
even if hardware ,software or any system management failure is there.
what is single point of failure?
1. node
2. power source
3. network adapter
4. network
5. disk adapter
WHAT TYPE OF FAILURE HACMP MONITORS
1. NODE and OS failure by using the redundant nodes
2. NIC failure by using redundant NIC cards
3. protection against network failure by using redundant network.
SHARED EXTERNAL DISK DRIVES:
A shared external disk drive is a disk connected to multiple nodes.
HACMP supports the following type of access to shared disk.
Non-concurrent access: only one node can access the data at one time. it uses serial environment.
if the node that is using the disk fails then failover happens
and other node becomes the active node and accesses the data.
concurrent access: the shared disks are actively connected to more than one node and data
can be accesses simultaneously . in this case parellel environment is used
and there is no need of failover.
A hacmp cluster comprises of
1. topology (nodes, NIC, network etc)
2. resources (things need to make highly available)
HACMP Topology components:
1. nodes
2. networks
3. communication devices
4. communication interfaces
what the term node refer in HACMP?
A node is a standalone server having hacmp software and member of the cluster.
In a HACMP cluster 32 nodes can be configured.
Network
In HACMP two types of network is defined.
1 . IP-Network
2. Non-IP network
IP Network uses the TCP/IP protocol suite for communication between the nodes.
NON-IP network is used for monitoring the status of the cluster and is strongly recommended.
This is because if there is network failure ,and only IP-network is used , it is diffcult for the active node to figure out whether it is node or network failure and this can lead to confusion and leads to cluster partitioning or node isolation. .
Non-ip network is used for differentiating between node and network failure.
Communication Interface:
In HACMP environment ,we are calling each Network interface which is having a ip_address and an ip-label associated with it. on the nodes which uses IP network for communication . we call
en0 (network interface)
ip-address (192.168.10.4)
ip-label (abhi_boot1)
Communication devices:
In HACMP environment , the devices used to configure the NON-IP Network is called communication devices.
it can be RS232( /dev/tty) device , disk (/dev/hdisk#) etc
communication devices provide point to point serial connection normally used for heartbeating .
What does resources means in HACMP?
As per my understandings, HACMP is completely designed for making the application highly availabilite.
So the things required to make the application run on the node is considered as HACMP resources.
take and example of application , what are the things required for making it highly available:
1. how to control(start/stop) the application : ( when some hardware error accurs ,resources need to move from one node to another , then hacmp must be aware how to start and stop the application that's why we are providing start/stop script )
2. how to access the application(to access you need ip and can move from one node to another )
3. common storage is required(if one node is down then other can access)
4. sometime nfs is required
Resouces can be:
1. application server
2. service- ip
3. volume group
4. filesystem
5. NFS mounts:
6. NFS exports
Application server : The application that has to be made highly available along with it's start/stop script is termed as application server in hacmp environment.
Service-ip/label: every application should have an IP address assigned to access it and that ip should be highly available
volume group: if the application requires shared storage, this storage should be coming from this VG .
filesystem : filesystems required for making the application highly available
NFS mount : NFS filesystems required for making the application highly available
NFS exports: NFS filesystem that need to be exported for running the application
What is Resource Group ?
Resource group is a collection all the resouces that are needed to make and application highly available.
while defining resouce group ,we need to specify polices(startup/failover/fallover) that controls the resource group action when it detects any failure.
As earlier , i have written that resource group is controlled by the policies.
the RG policies are :
1 Startup policy (used when cluster starts up.)
2. Fallover policy (if node failure is there, it determines which node to takeover)
3. Fallback policies (when the higher priority node comes up, then this policy decides whether to
move the RG to higher priority node or node)
startup policy:
online on home node only : The resource group is brought online only on its home (highest priority
HACMP stands for High Availability Cluster Multiprocessing
Hacmp provides two types of environment
1. serial (high availability): this is used to make the application highlt available
by using the shared disk and duplicate resources . also the access to data will be serially
means one node at a time.
2. parallel (cluster multiprocessing ) : this type of environment is used when the application is
online on many nodes and allthe nodes can concurrently access the data. for this type of environment
failover is not required.
why to use hacmp?
1. eliminating single point of failure
2. elminating planned and unplanned downtime.
hacmp ensures that the application should be highly available and accessible
even if hardware ,software or any system management failure is there.
what is single point of failure?
1. node
2. power source
3. network adapter
4. network
5. disk adapter
WHAT TYPE OF FAILURE HACMP MONITORS
1. NODE and OS failure by using the redundant nodes
2. NIC failure by using redundant NIC cards
3. protection against network failure by using redundant network.
SHARED EXTERNAL DISK DRIVES:
A shared external disk drive is a disk connected to multiple nodes.
HACMP supports the following type of access to shared disk.
Non-concurrent access: only one node can access the data at one time. it uses serial environment.
if the node that is using the disk fails then failover happens
and other node becomes the active node and accesses the data.
concurrent access: the shared disks are actively connected to more than one node and data
can be accesses simultaneously . in this case parellel environment is used
and there is no need of failover.
A hacmp cluster comprises of
1. topology (nodes, NIC, network etc)
2. resources (things need to make highly available)
HACMP Topology components:
1. nodes
2. networks
3. communication devices
4. communication interfaces
what the term node refer in HACMP?
A node is a standalone server having hacmp software and member of the cluster.
In a HACMP cluster 32 nodes can be configured.
Network
In HACMP two types of network is defined.
1 . IP-Network
2. Non-IP network
IP Network uses the TCP/IP protocol suite for communication between the nodes.
NON-IP network is used for monitoring the status of the cluster and is strongly recommended.
This is because if there is network failure ,and only IP-network is used , it is diffcult for the active node to figure out whether it is node or network failure and this can lead to confusion and leads to cluster partitioning or node isolation. .
Non-ip network is used for differentiating between node and network failure.
Communication Interface:
In HACMP environment ,we are calling each Network interface which is having a ip_address and an ip-label associated with it. on the nodes which uses IP network for communication . we call
en0 (network interface)
ip-address (192.168.10.4)
ip-label (abhi_boot1)
Communication devices:
In HACMP environment , the devices used to configure the NON-IP Network is called communication devices.
it can be RS232( /dev/tty) device , disk (/dev/hdisk#) etc
communication devices provide point to point serial connection normally used for heartbeating .
What does resources means in HACMP?
As per my understandings, HACMP is completely designed for making the application highly availabilite.
So the things required to make the application run on the node is considered as HACMP resources.
take and example of application , what are the things required for making it highly available:
1. how to control(start/stop) the application : ( when some hardware error accurs ,resources need to move from one node to another , then hacmp must be aware how to start and stop the application that's why we are providing start/stop script )
2. how to access the application(to access you need ip and can move from one node to another )
3. common storage is required(if one node is down then other can access)
4. sometime nfs is required
Resouces can be:
1. application server
2. service- ip
3. volume group
4. filesystem
5. NFS mounts:
6. NFS exports
Application server : The application that has to be made highly available along with it's start/stop script is termed as application server in hacmp environment.
Service-ip/label: every application should have an IP address assigned to access it and that ip should be highly available
volume group: if the application requires shared storage, this storage should be coming from this VG .
filesystem : filesystems required for making the application highly available
NFS mount : NFS filesystems required for making the application highly available
NFS exports: NFS filesystem that need to be exported for running the application
What is Resource Group ?
Resource group is a collection all the resouces that are needed to make and application highly available.
while defining resouce group ,we need to specify polices(startup/failover/fallover) that controls the resource group action when it detects any failure.
As earlier , i have written that resource group is controlled by the policies.
the RG policies are :
1 Startup policy (used when cluster starts up.)
2. Fallover policy (if node failure is there, it determines which node to takeover)
3. Fallback policies (when the higher priority node comes up, then this policy decides whether to
move the RG to higher priority node or node)
startup policy:
online on home node only : The resource group is brought online only on its home (highest priority
node defined ).if it is unavailable RG will not start automatically.
available on first available node : The resource group comes
online on the first participating node that
becomes available.it's just like rotating RG.
using distribution policy : here only one RG is active on one node. if one node is having an active RG
,it will goto second if it is not having any active RG.
on all available nodes: This policy is used when you need that all nodes can access the same RGconcurrently.
Fallover/failover policy:
Fallover to next priority node in the list : at the time of failure,RG is moved to the next high priority node in
the RG definition
Fallover using dynamic node priority : if using this option, then the fallover will happen according to the
selected criteria from the below methods
1. highest_mem_free
2. highext_idle_cpu
3. lowest_disk_busy
Bring offline(on error node) : this is used with concurrent RG.since RG is active on all nodes concurrently,
if one node goes down, then RG on this node should become online for the maintenance of that node..
Fallback policy
fallback to higher priority node in the list: when the highest priority node comes up, the RG is
automatically moved to it.
Never fallback : it will not fallback to higher priority node automatically, you
need to move the RG manually.
--------------under construction-----------------------------------------
