Hello,

since april last year we get error 510, at the moment once every day, always about 10:30am. Then we stop and start BaaN and oracle, so we have no problems for the rest of day.
See above the message from event viewer:

*******************************************
Env BaanIVc3 (D:\baan)
Prog ora7_srv file \db\servers\ORACLE_2\ora_native.c # 1867
Keyword Oracle Error
Username wru type N language 3
Process 0x732
Lasterror 0
Errno 0
bdberrno 0
Message
dbs_errno = 0 Error 510 occurred:
Error ORA-12203 occurred during logon.
ORA-12203: TNS:unable to connect to destination

Error BDB-510 returned.
Check the Oracle settings:
NLS_LANG = '' (default US7ASCII)
ORACLE_HOME = 'd:/orant'
Oracle Service Name = '' (resource oracle_service_name --> LOCAL)
ORACLE_SID = 'BAAN'
Logon 'wru' as 'wru' failed; errno 0
Flushed at \db\servers\ORACLE_2\ora_driver.c : #203.

*********************************************

We've got an Windows NT 4.0 Server with service pack 6a.
4x pentium PIII Xeon 550MHz processors
4 GB RAM
BaaN IVc4
120 users
porting set 6.1c.05.02
Oracle 7.3.4.2.1 - Level 2 driver
Reboot every night

Attached you can find the bd_resource and initbaan.ora

Is there something wrong in our environment. No one can help us, not an oracle specialist not baan support.

Thanks Ingo

Have you checked the Oracle log files (alert log) for anything strange happening with the database? If the problem always occurs at exactly the same time then it sounds like some sort of scheduled job is causing this.

No entry in oracle alert.
It's not always exactly the same time, but normaly between 10:00 and 11:00am. There are no jobs at that time - we checked.
Perhaps queries, we asked the users, but no reply until now

Regards
Ingo

challenging!

1) Have you experienced days without this behaviour or is it EVERY day?
2) What about the workload on the NT-Server at the time of crash?
3) You said that you "stop and start Baan and Oracle" - so both services are up and running? Can you connect to oracle at the time of crash if you use something different than Baan?
4) NT-Server "scheduled tasks" is empty?
5) Any Anti-Virus Software running on the Server?
6) What if you change the system time (let's say +2 hours) - does the time of the crash change?
7) What about the schedules of your backup-SW?
8) Any Baan-Jobs scheduled?
9) $BSE/log ... any files?
10) NT event log ... nothing?


Sorry if some questions seem to offend you, this is not the intention, I just wanted to write everything down that I would look for.

Keep us informed - this is interesting!!

(About 15 years ago a similar case was solved only when someone sat beside the "server" the whole day and found out that the sun being reflected by the automatic shutters of the neighbour building's windows overheated the system *grin*)

Good luck

Norbert

Another potential thing to check: run the TNS Listener in 'debug' mode for a while and check those log files when the incident happens.

Dear Norbert,

here are my answers:

1) Have you experienced days without this behaviour or is it EVERY day?
=> no error at weekend and last two weeks at monday. So we just wait till 10:00 this day (monday). It is Monday 9:36am and error 510

2) What about the workload on the NT-Server at the time of crash?
=> about 3,3 GB (4 GB RAM and no NT Enterprise Edition)

3) You said that you "stop and start Baan and Oracle" - so both services are up and running? Can you connect to oracle at the time of crash if you use something different than Baan?
=> most time we could connect with sqlplus. I try it next time. Its is possible to connect with sqlplus

4) NT-Server "scheduled tasks" is empty?
=> No. We use scheduler for terminating BaaN Jobs.

5) Any Anti-Virus Software running on the Server?
=>Yes - McAfee. But no online scanning, only terminated scan at the evening

6) What if you change the system time (let's say +2 hours) - does the time of the crash change?
=> Good idea. We check this.

7) What about the schedules of your backup-SW?
=> We use Arcserve. No jobs at that time.

8) Any Baan-Jobs scheduled?
=> Not at that time

9) $BSE/log ... any files?
=> no

10) NT event log ... nothing?
=> messages like this:
*********************
Env BaanIVc3 (D:\baan)
Prog ora7_srv file \db\servers\ORACLE_2\ora_native.c # 1867
Keyword Oracle Error
Username wru type N language 3
Process 0x732
Lasterror 0
Errno 0
bdberrno 0
Message
dbs_errno = 0 Error 510 occurred:
Error ORA-12203 occurred during logon.
ORA-12203: TNS:unable to connect to destination

Error BDB-510 returned.
Check the Oracle settings:
NLS_LANG = '' (default US7ASCII)
ORACLE_HOME = 'd:/orant'
Oracle Service Name = '' (resource oracle_service_name --> LOCAL)
ORACLE_SID = 'BAAN'
Logon 'wru' as 'wru' failed; errno 0
Flushed at \db\servers\ORACLE_2\ora_driver.c : #203.
*******************


Dear Patrick,

on wednesday is an oracle specialist in our office. I ask him to start Listener in debug mode.

Best regards
Ingo

Dear Ingo,

have you ever had the nerve/time to wait whether the system continues to work after a period of time?

==> 4) "terminate Baan Jobs" via NT scheduled jobs
Have you tried to disable these jobs?

Please keep us informed.

regards

Norbert

Dear Norbert,

sometimes, but not often, there is an error 510 only a few seconds and then system is ok. But normaly the situation increase, becomes more and more worst. So that it isn't possible for the users to work. You mean waiting can help? But you know that enduser can't wait - for them 5 min. are too long.
The use of windows nt scheduler is historical, because at (our) beginning baan job-deamon was buggy. Is it now ok?
But as I told before there are no jobs at that moment.

Best regards
Ingo

A very long time ago I experienced an error 510 (I believe with a new installation). There it was due to an error in the tabledef file.

At that time we had c:orant instead of c:\orant
I see your oracle_home is with a forward slash (/) and not a backslash (\).

But this is just a (stupid ?) shot in the dark.

Hi Norbert,

yesterday we got error 510 twice. First at about 9:50 and again at 15:20. The second we just wait a few minutes (about15 min) then situation becomes better.

Sorry for my english:

4) NT-Server "scheduled tasks" is empty?
=> No. We use scheduler for terminating BaaN Jobs

I mean not killing jobs, but starting them with at-command.

Regards
Ingo

Dear Ingo,
Could you please tell me, you are restarting your server or database daily?
As my experienced regarding ORA-12203, it may occurres in case your database server is too slow and cannot communicate with TNS, this may happen eventually.
If the answer is yes for above questions, following things are happening at your side.
The moment you restart server, all the BaaN services are getting available to users but for opening database it is taking some time because reading services, mounting database and opening database precesses are time consuming (Depends on database size and locations for datafiles) and after that removing data from Temp. Tablespace oracle uses lot of resources, result your server goes slow-down and you are getting TNS listener error.

For solving your problem, you can try:
1. Avoid restarting server/database daily.
2. If not possible 1st issue, deliver system after proper starting oracle. (you can disable/enable users.
3. Observe reason for increasing heavy Temp. Tablespace.( May be wrong queries)
4. Observe your server in no user mode (for users Job)

Do reply, so that I can go in depth.

-Suhas

Hi,

yesterday was an oracle specialist in our office. He checked some things. As a result we can say: error 510 is coming when there are more then 900 sessions in oracle
=> select count(*) from v$session;

When we check the number of sessions by user:
=> select username , count(*) from v$session
group by username order by count(*);

We get a lot of users with more then 20, some with more then 30 sessions. That seems to be the reason. The quantity of sessions increase the amount of used meomory. And with nt 4.0 and oracle 7.3 we only can use 2GB. So in future we restrict the number of process of each user in baan. Most user will say that they can't work anymore. But doesn't matter. The next thing I try is an new porting set (6.1c.06.03) on our test server and new GUI (063) on client. With this I can use 'session_timeout'. We are thinking about it. This function will close the complete bshell - is it so?
Then we must give some users the option to open some sessions with bshell start. At the moment I'm trying utility 'CloseIdle'. I read that this tool can close individual sessions, but until now it don't work correct.

How do you manage this problem of quantity of sessions?
Which restrictions do you use?
Is there a possibility to disconnect sessions after a certain time but not close them, so that a user just by clicking can reconnect them?

Regards and thanks a lot
Ingo

In the baan session Maintain user data form 3 you can set the Max number of processes.
We limit this to 10 for most of the users.

You can put SESSION_TIMEOUT to kill "old" idle sessions. Put f.e. 150 mins. I think you don't need CloseIdle. Many customers wants to open sessions beforehand. If they are using Menu browser, introduce worktop interface for users instead! Users can create shortcuts of sessions quickly.
We have had those 510 errors in unix world many times. Check your user data that there is no limits with processes.
If session parameter is a bottleneck, there will be ORA error in log!

We use a limit of 5 per user. No complaints in 3 years.

Dirk

Hi Ingo,

problem solved or not?

-Suhas

Hi Suhas,

see my thread from 27. March. At the moment we can control the situation and users don't open as much sessions as before. For users who won't work with us, we're going to use session_limit

Regards Ingo