Datasheet
“main” (Installation and Administration) — 2004/6/25 — 13:29 — page 397 — #423
i
i
i
i
i
i
i
i
19
High Availability under Linux
take over the address and task of the failed component. Redun-
dancy expressly relates to the two interface directions. There is
no harm planning an active and backup interface for both the
internal and external interfaces.
Hard Disks Assign several hard disks to your system and arrange
the data backup (e.g., using RAID or drbd) in such a way that
if one of these disks is lost, the others always contain the intact
data record. It must be possible to replace a faulty disk with a
new one without stopping the system.
Applications All important data and applications that form the outer face
of your systems must be arranged in such a way that they will not
prevent a restart. If an application does not release its lock files after a
crash, this prevents the relevant process from restarting. This means
that the application is not suitable for a high availability environment.
Ideally, the “health” of certain applications, operating system pro-
cesses, and network connections should be monitored with a suitable
monitoring tool.
Data After a system fails, all key data must be available to the failover sys-
tem complete and intact. This type of high availability is achieved
by distributing stored data over several systems or hard disks. For
this, the contents of a disk are regularly mirrored on another disk (or
several disks), which can take over with the intact data record if a fail-
ure occurs. Use a journaling file system to ensure that a file system
restarts in a consistent state after a system crash.
Network All network infrastructure should be configured for redundancy,
from the router and switch infrastructure down to the simple net-
work cable.
19.4 The Software Side of High
Availability
The key software aspects of high availability solutions are described below.
19.4.1 heartbeat
heartbeat is a package that is used to monitor all the nodes used in the clus-
ter. heartbeat exchanges “heartbeats” on the network interfaces of the
397SUSE LINUX Enterprise Server