Spectrum Online Backup taking a long time to complete in a Fault Tolerant environment

Document ID : KB000125097
Last Modified Date : 22/01/2019
Show Technical Document Details
Issue:
Spectrum Online Backup taking a long time to complete in a Fault Tolerant environment installed on Linux
Environment:
Spectrum 10.x Fault Tolerant installed on Linux
Cause:
A check of the RCPD.OUT files on both the primary and secondary SpectroSERVER systems shows the file transfer of the backup file from the primary to the secondary is taking a long time. The following is an example:
 
Jan 15 09:25:04 : Starting file transfer using 1048576 byte TCP receive socket buffer.
Jan 15 10:31:00 : Successfully received .ft_save_file.SSdb.gz
Resolution:
Increase the OS TCP socket buffers on both the primary and secondary SpectroSERVER systems:

1. Log into the SpectroSERVER system as the root user
2. Edit the /etc/sysctl.conf file and add the following:

net.core.wmem_max=1048576 
net.core.rmem_max=1048576 
net.ipv4.tcp_rmem= 10240 87380 1048576 
net.ipv4.tcp_wmem= 10240 87380 1048576


3. Save the changes to the /etc/sysctl.conf file

4. Run the following command as the root user:

sysctl -p

Increase the TCP buffer used by the rcpd program on both the primary and secondary SpectroSERVER systems:

1. Log into the SpectroSERVER system as the user that owns the Spectrum installation

2. cd to the $SPECROOT/SS directory

3. Edit the .rcpdrc file if it exists. If it does not exist, create new

4. Add the following to the .rcpdrc file:

tcp_buffer_size=4194304 
send_buffer_size=4194304


5. Save the change to the .rcpdrc file