Coyote Point Systems Equalizer Especificações descarregar pdf (Página 424)

Failover

Understanding Failover

In an Active/Passive failover configuration, two Equalizers are configured into active and passive roles, with the

active Equalizer serving cluster traffic. A "failover" is said to occur when the active Equalizer stops processing

client requests and the passive Equalizer starts processing cluster traffic.

When two Equalizers are configured into this failover configuration, they form a "failover pair". An Equalizer in a

failover pair is called a "peer". At any given time, only one of the Equalizers in a failover pair is actually servicing

requests sent to the cluster IP addresses defined in the configuration -- this unit is called the "active peer" or the

"current primary" Equalizer in the pairing. The other Equalizer, called the “passive peer” or "current backup", does

not process any client requests.

Both units continually send "heartbeat probes" or "failover probes" to one another. If the current primary does not

respond to heartbeat probes, a failover occurs. In this scenario the current backup Equalizer assumes the primary

role by assigning the cluster IP addresses to its network interfaces and begins processing cluster traffic.

In an Active/Active failover configuration, clusters are Active on

both

peers in the configuration. For the same

failure situations that cause a peer to take over all the cluster and floating IP addresses in an Active/Passive

failover configuration, Active/Active failover operates the same way - that is that a healthy peer will take over all of

the cluster and failover IPs. This type of failover configuration is available when using two Equalizers using EQ/OS

10.

This section provides instructions for the configuration of Active/Passive failover between an EQ/OS 8.6 and

EQ/OS 10 system and two EQ/OS 10 systems and Active/Active failover between two EQ/OS 10 systems.

How Equalizer Determines if it Should Assume the

Primary Role

Equalizer expects to see a heartbeat from a failover peer (all failover peers in Active/Active Failover configuration

[See "Configuring Active/Active Failover Between Two EQ/OS 10 Systems" on page 456]).within a "heartbeat

interval" (time in seconds) on every subnet where a heartbeat flag is configured. When a "Failed Probe Count"

value is reached – that is, when that number of heartbeat probe intervals has elapsed without receiving a heartbeat

from a peer -the backup unit will attempt to assume primary role.

Equalizer will:

1. Check to see if any of the cluster or failover IP addresses defined in its configuration are already on the

network -using ICMP requests/replies.

2. Perform checks on its locally defined networks and determine the number of subnets on which it has

connectivity. It compares this information to the connectivity information in the last heartbeat received from

the peer that reached the "Failed Probe Count" value.

a. If the Global "Failed Probe Count" on the failover configuration = 0, then the "Failed Probe

Count" configured on the subnet will be used to determine when failover occurs.

b. If the Global "Failed Probe Count" is reached BEFORE the "Failed Probe Count" configured

on the subnet, then failover will occur.

424

1 2 ... 419 420 421 422 423 424 425 426 427 428 429 ... 593 594

Sem comentários

Coyote Point Systems Equalizer Especificações Página 424