Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> RE: Oracle networking question

RE: Oracle networking question

From: Michael Ray <Michael.Ray_at_trw.com>
Date: Wed, 22 Nov 2000 06:44:33 -0800
Message-Id: <10688.122734@fatcity.com>


Tim,
I would really be interested to know if you think that the discconect/recon= nect every day "solves" your problem. While the great majority of our apps = are written in VB, they connect via ODBC not OO4O. However, our main = contractor uses OO4O for a DCOM server which all his apps connect through. = Since these are just OLE objects, it isn't something I can just stop and = restart to see if that would fix it if this ever occurs again. I would = have to kill all the processes. Anyway, this DCOM server has been around = for quite some time and this was the first time I had experienced this.

I don't think that was the issue though since moving files is the only = thing we did.

TNS_ADMIN has never been set on that server. The files were in the default = location. So apparently, NT can lose Oracle registry values too??? Putting = the files in the Bin directly solved it since NT always looks in the app = directory first as you noted. The Network/Admin directory is not put in = the path so the executables either know to look their internally or it is = set in the registry somewhere. But why would that stop working out of the = blue?

Michael Ray
Oracle DBA
TRW, Marshall, IL
217-826-3011 x2438

>>> tim.onions_at_speechmachines.com 11/22/00 03:41AM >>>
Michael

I have experienced something very similar on NT but with a different solution. We found that after between 2-4 weeks of continued successful running Oracle 8.1.6.2.0 would suddenly not allow any new connections = (which
simply hung and never came back with an error nor ever timed-out). = Listener
was running OK. I spent ages on the phone with WWS about this and they = could
not come up with a solution other than stopping/restarting the instance = (no
mention from them about moving the TNS files). The funny thing was that in preparation for the shutdown we started disconnecting application = processes
(which are multithreaded C++ with OO4O by the way) and as soon as we = stopped
a certain one of these all new connections were magically allowed once again.

The same problem occurred a couple of weeks later and was fixed in the = same
way - disconnecting a certain application OO4O set of processes.

We now have our OO4O processes automatically disconnect/reconnect every 24 hours but it is too early to say whether this has fixed the problem yet.

Are there any similarities inthe above with your set-up?

Another thing we have experienced is that NT has been know to "lose" environment variables - when it does so it cannot find certain products. This, for us, means that Net8 cannot find TNS_ADMIN environment variable = and
so cannot connect. I would guess (but have not tried) that moving the TNS files to the \bin directory would fix this an NT executable will look in = Received on Wed Nov 22 2000 - 08:44:33 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US