tcp_ip_abort_interval and Service Guard

[Sajith V Mannadiar]

Hello there,

We have an Oracle Application server running on HPUX 11i and this node is a Single Node cluster that has the application server configured in MC/SG packages for monitoring & High Availability

This server connects to one of the four LDAP servers available.

We would like to change the tcp_ip_abort_interval parameter to 60,000 (1 minutes) to allow the application server to fail-over to an alternate LDAP server in the event of the primary not being reachable

With the default value of tcp_ip_abort_interval (which is 600,000 or 10 minutes), it takes almost 10 minutes before it decides to fail-over.

Our concern is, will this have any other impact (like with Service Guard or the database connections)?

Please advice..

Thanks,
Sajith

[Ron Kinner · Accepted solution]

Sun uses a 1 or 2 minute interval so it should work unless you are dealing with a mail server. Look at the recommendations on page 16 of 90 in
http://edocs.bea.com/wls/docs70/pdf/perform.pdf

or better since it's from HP (page 8 of 9)

http://www.hp.com/products1/unix/webservers/apache/infolibrary/pa_hpapache_admin_guide.pdf

Don't think it should have an effect on Service Guard.

Ron