The Question is:
We have occasionally experienced a "hang" when issuing a tcpip "connect" via
sys$qiow(tcpipChan, IO$_ACCESS, &iosb, .....)
This has happens when the remote host has "a problem" (it is a vendor supplied
windoze XP system and sometimes the vendor's application goes wild or
something). But what ever the case, shouldn't the qiow return with an error
at some point instead of hang
ing? Are they any patches for this behavior? Or must I use a qio version and
cancel the i/o after waiting for a specified time?
Thanks for any advice,
The Answer is :
Please contact the support center, as details of the TCP/IP version
and ECO (V5.4 with ECO being current, as of this writing), as well
as details of the hang (preferably leading to a way to reproduce
the configuration), will be required. This can discussion can and
likely should also involve the third-party vendor, obviously.
Depending on what is actually occuring on the remote end of the
connection, the connection might well appear to hang.
You can certainly issue cancellation requests, but these can
variously be as problematic as the hang itself -- applications
and networks in indeterminant states tend not to cancel
particularly cleanly. An asynchronous call and a timer, or
a synchronous call and an AST on a timer can potentially be
used to cancel the operation.