Anyone run TCP_TUNE.SH ?

[S_Williamson]

Under PD 6.61 tuning guide, there is a way you can tune your replications . I was wondering if anyone has attempted this? We have a few sites with large amounts of changes which are falling behind so was looking into tuning them. You need to run the script on Source & Dest which means running across around 6 sites for me so didnt want to do this without some advice. I may end up logging a call for it. Thanks.

# /opt/pdconfigure/scripts/support/tcp_tune.sh
Version: 0.1
Usage: tcp_tune.sh {modify|restore|show}
where,
     modify  will set the default maximums to 4MB
     restore will restore the default values
     show    will show the current settings

This script will modify the read and write buffer TCP
settings tcp_rmem and tcp_wmem to use 4MB as the default max
Note: run this script on the source and destination

# /opt/pdconfigure/scripts/support/tcp_tune.sh show
Version: 0.1

/etc/sysctl.conf

/proc/sys/net/ipv4/tcp_rmem
4096    87380   174760
/proc/sys/net/ipv4/tcp_wmem
4096    16384   131072

[Jzombory]

Support recommended that run the tcp_tune.sh script. We have a very large environment and have had no problems with local backup jobs or replication jobs before or after running the script.

I see no down side to running this on all your nodes.

[Douglas_Snyder]

Hello,

The script is the supported method for tuning replication performance.  As with any changes to your infrastructure it is recommended to log a help disk ticket prior to implementation.

BTW - I have run the script both in test and production environments with no problems.

[patricpaal]

Hi,

I'd like to run the tuning script in our ENV but I need to know if the PD server requires a reboot after running the script.

Thanks