Discussion Locked

NW6 unable to clear connections filelock

0 helpful
13 replies
Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

Hello all,

Issue: NW6 Server DS ver 8.6.2
Users: Lose connection to server or lockup when using database applications. They relogin to server try to open files and Error : Cannot open in use exclusively by THEMSELVES even though sharing is enabled at database and file level and all FS rights equal!
Findings: User has multiple connections open but only using one. Other connection has files locked. Cannot clear connection through Monitor or through Console(server locks up at console).
Files Cannot be deleted moved copied etc. Workstations running W98/NT/2000 client ver 3.1 sp2, 3.2 sp1, 4.83 sp1. Server has been up without incident since 8/02

13 replies

by nicholasyonko

This is a problem I have found with some 3com NICs - specifically 3C905's. What are your workstations running?

by malibu81

3c920 integrated FE - I Should have noted all networks segements have been sniffed for basic framing and packet errors, drops and losses. FK

by deroode

It may be possible to clear the connections using the management portal. This will clear the connection, but not release the lock. The locks can be released by dismounting and mounting the volume.

The cause may be a bad behaving lan driver, disk driver or Antivirus software. Try disabling antivirus first, upgrade the lan driver on the server, maybe even swap the nic for another type.

Make a baseline performance chart of your processor utilization, especially from interrupts. When the problem manifests look again at processor utilization, and see where more work is being done. Suspect that device.

by malibu81

Antivirus was rolled back to previous definitions, NG. It was unloaded. NG. Lan Driver replaced current and later rev. on board NIC swapped for NEW replacement NG. PROCESSOR UTILIZATION normal all throughout incident. - FK

by tommymcbrayer

I know that Novell new about this in Netware 5 but I thought they had fixed this in 6. May I ask what kind of database this is? IE some type of SQL or BTRIEVE yadda yadda ya. You say that the users are losing connection to the database. Are the users actually running the program and the system seems like it locks up or becomes unresponsive and they force disconnect from the app. This was a major pain for us in 5. User was in a database and it would be processing a request but not fast enough for the user so they would terminate the app. Is this happening with all users or just a select few? Once I have more info I can follow up more clearly.

TL McBrayer CNE6, CNE5, CCNA, CXE

by malibu81

ACCESS 97 ALPHA4 SQL QRACLE ACT - Only Rejected because this is a request For More Information FK

by tommymcbrayer

Cool, answer try
set client file caching enabled = off at the server console and see if this corrects your problem. This is a good thing on a database server. Do you get any server errors or just mulitple connections in monitor, what about a false high processor utilzation in monitor or is it the database locks, user tries to reconnect but can't becasue their old connection still is running the process. Pain, but this is designed in the NOS by default to try and prevent database corruption. Turning th client file cacheing off on the server should fix you up. Been there done this.

TL McBrayer CNE6, CNE5, CCNA, CXE

by malibu81

We tried set file caching enabled = off NG. added file cache level = 0 in client NG. - OTHER IDEAS
Seems multiple connections and more apparent with ACCESS 97 and ALPHA4 - FK

by malibu81

Users: Losing drive connectivity on conection a#
while in database applications only.

Users: Then reconnect to erver on connection number b#

Users: Unable to reconnect because connection a# still has files locked.

Admin: Is able to close file lock but not connection a#

- FK

by tommymcbrayer

Hmmmm, thought that might have done it. Have you turned it off at the client as well. I would also look at

For systems with heavy Access 97/2000 users, increase the MAXIMUM RECORD LOCKS PER CONNECTION to 10000 and the MAXIMUM RECORD LOCKS to a higher level commensurate with users and number of concurrent Access queries.

For example, if you have 10 concurrent access users and 100 users total on the system, setting this parameter to 150000 is a good starting point).

If this no worky Iwould do the normal and ensure sp2 loaded on server and the latest client on workstaion. If none of these prevail, I would contact the vendor to address your database files being locked. (client connections can't clear).

TL McBrayer CNE6, CNE5, CCNA, CXE

by malibu81

We allready had the record locks set - It was a combination of several things 1. Client file caching fixed at sever and client (even though this raised utilization by X2) it helped. 2. Rollback of NLSLSP.NLM to origional shipped version versus that supplied by SPK2. 3. Latest Client was applied to WIN 95/98 users Version 3.3.2 spk 1 4. All seemed cool for a while but it croped up again. After sending out my PC Guy's it was determined that 2 heavy Database users wer trying to gain exclusive access to datasets over 2 GB. The machines were so underpowered that they came to a halt. The server seen the delay as a disconnect, locked the files and tried to reconnect. User rebooted and connection at server was remaining active. New NLSLSP.NLM fixed the locking clearing problem of lisc and connections. Caching settings fixed the majority of issues with all clients. Hardware upgrade and proper database maint. (integrity checks, repair and compacting from my DBA's will insure speed and no wasted RAM)- GOOD INSIGHT & THANKS FK - CNE5, CNA4x,3x,2x MPE ADMIN UX ADMIN REMEMBER THE BIRD (LINUX RULES)

This discussion is closed to new replies.