Appendix C: Active/Active High Availability Support
TIBCO MFT Platform Server for UNIX now supports Active/Active High Availability (HA) when running behind a load balancer. This allows multiple systems to appear to the transfer partner as a single system. The following figure illustrates how high availability works:
-
Platform Server clients, Internet Server clients, and Command Center administrative functions connect to a load balancer.
-
The load balancer connects the client to any of the Platform Server instances in the cluster.
-
The Platform Server instance contacts the Primary
CyberMgr
to get transaction numbers and write audit logs. If the PrimaryCyberMgr
is not running, then it contacts the SecondaryCyberMgr
. -
Platform Server can directly read the configuration files from a common location accessible to all instances in the cluster.
-
Transfers can execute on any of the instances in the cluster.
-
Platform Transfer restarts can execute on any of the instances in the cluster.
-
Audit inquiry can execute on any of the instances in the cluster.
-
Transfers initiated by HA cluster Platform Server Clients have not changed, except that they use CyberMgr to get transaction numbers and write audit records.
These are the goals of the Platform Server HA feature:
-
Multiple (two or more) PSU servers appear to be a single PSU Server
-
Share configuration files
-
Unique transaction numbers
-
Support checkpoint Restart
-
No changes to PS Client User Experience
-
Share Log.txt (i.e. Transfer Audit) files
-
BW-initiated transfers can poll target servers for transfer completion
-
Clients can connect to any Platform Server in the cluster
-
Command Center Collection works seamlessly even when connecting to different Platform Servers.
Key Components of Platform Server HA
Single location for files and data shared by multiple server in the cluster are as follows:
-
Config files, (except for config.txt) nodes, and profiles
-
Log.txt files
-
Transaction Number file
-
PQF files
-
Audit file inquiry (cfinq or Command Center initiated)
-
Active transfer inquiry
CyberMgr RPC Server handles all access to configuration and log files. The primary and secondary CyberMgr
servers provide failover in case of a failure.
Key Features of CyberMgr
CyberMgr
is started on each PSU Instance. It writes to all the files that need single threading. These files are: Transaction Number
, Log.txt
, and Messages.txt
.
CyberMgr
is used for both HA and non-HA installations. CyberMgr
runs with the same authority as CyberResp
, which is root, when CyberResp
runs as root.
The config.txt
parameter that defines the TCP Port CyberMgr listens on CyberMgrPortLocal: 46678
.
Transfers fail if both primary and secondary CyberMgr
are down.
CyberMgr performs the following tasks:
-
Gets Transaction Number
-
Writes Log.txt
-
Writes message log (always done by Local
CyberMgr
) -
Writes audit log (always done by Local
CyberMgr
)