
no checkpoint session periodic-interval
end
Selective Disabling of NACK Messaging
The NACK mechanism sends a NACK message for any ICSR checkpoint failure on the standby chassis.
Every NACK sent from the standby chassis triggers a full checkpoint from the active chassis.
If the micro-checkpoint is failing continuously and sending NACKs, the active chassis keeps sending
full-checkpoints. This increases SRP bandwidth.
CLI commands allow an operator to selectively disable and re-enable NACK messages for specific
micro-checkpoints.
The Exec mode
srp disable nack micro-chkpt-cmd
disables the sending of a NACK from the standby chassis.
srp disable nack micro-chkpt-cmd chkpt_number
chkpt_number
specifies the checkpoint number to be disabled as an integer from 1 through 255. You can
obtain checkpoint numbers (CMD IDs) from the output of the
show srp checkpoint info
command.
You can re-enable the micro-checkpoint using the
srp enable nack micro-chkpt-cmd
command.
srp enable nack micro-chkpt-cmd chkpt_number
Configuring LZ4 Compression Algorithm
You can optionally enable LZ4 compression algorithm for SRP messaging payload. The zlib algorithm remains
as the default.
LZ4 is a very fast lossless compression algorithm with near-linear scalability for multi-threaded applications.
The
compression
keyword in the SRP Configuration mode
checkpoint session
command allows you to
enable the use of the LZ4 compression algorithm.
The
compression
keyword will only appear if a special ICSR optimization feature license has been
purchased and installed. Contact your Cisco account representative for assistance.
Important
The following command sequence enables the use of LZ4 compression:
configure
context context_name
service-redundancy-protocol
checkpoint session compression lz4
end
LZ4 compression is effective only if both chassis are configured with LZ4. If any one chassis has zlib (default)
configured, the compression algorithm reverts to zlib. The algorithm is negotiated only during initial socket
establishment. Once agreed no more negotiation takes place until the TCP socket connection is reset.
Reducing Sync-Up Time with Standby ICSR Chassis
The default method for synchronizing the SRP database requires tens of seconds of delay whenever the TCP
connection between the Active and Standby session managers is established. Once the TCP connection is
established, heart beat messages are exchanged between both ICSR chassis every 3 seconds. The standby
ASR 5500 System Administration Guide, StarOS Release 21.5
281
Interchassis Session Recovery
Configuring the Service Redundancy Protocol (SRP) Context