Re: SQL*NET V1 Hangs!

From: Roderick Manalac <rmanalac_at_oracle.com>
Date: 20 Oct 1994 02:52:47 GMT
Message-ID: <384m1v$n96_at_dcsun4.us.oracle.com>


wkvanmet_at_cacd.rockwell.com (W K VAN METER) writes:
|> We were having a similar problem and found that specifying a timeout
|> for user processes using sql*net helped.
|> The command is:
|>
|> tcpctl timeout 5
|>
|> This forces a "hanging" process to timeout.

Actually it forces orasrv to give on waiting for a reply from a hanging connection attempt. The typical scenario would be a user on a PC who types 'sqlplus scott/tiger_at_t:host:SID' hits return and immediately reboots the PC.

Once orasrv does fork a dedicated server for the connection request, it's mostly out of the picture (waiting for a SIGCLD while processing other connection requests) and the timeout option is no longer in play for the process. In other words, it is dependent on the keepalive mechanism within TCP/IP at that point.

Roderick Manalac
Oracle Corporation Received on Thu Oct 20 1994 - 03:52:47 CET

Original text of this message