While a reload won't affect existing connections, it *does* have the
unfortunate side effect of discarding all dynamic service registrations...
behaving much more like a restart than you're likely expecting.So you can
expect to see periodic "ORA-12514: TNS:listener does not currently know of
service requested in connect descriptor" errors until the database
instances re-register. In a RAC configuration this change does *not*
propagate to the SCAN listener, so it will happily continue to direct
clients to the temporarily impaired node. This is true through at least
version 18.5... I haven't tested on a 19c environment yet.
For a non-clustered environments, you're going to want to plan for a brief
period of disrupted connectivity. It should be relatively simple to script
an "alter system register;" command for each DB, if desired, in order to
minimize this period.
For RAC databases you can simply stop all services on the local node
beforehand using *srvctl*, and then restart them after all of the DB
instances have re-registered with the listener.
On Mon, Oct 14, 2019 at 12:06 PM Rakesh Ra <rakeshra.tr@xxxxxxxxx> wrote:
I wanted to add sqlnet.inbound_ connect_ timeout = 300 in my sqlnet.ora
file (both grid and oracle home) and reload my listener ( running from
grid) without affecting application connections. As reload of listener just
reads the configuration files without restating the listener , I hope this
works , and even if it works where can I validate that listener has indeed
picked up the new value?