File Server Hangs Multiple Shares/Files Open
This is sort of a long shot but...
We recently built a new Windows 2003 R2 Standard server on an IBM server. Dual-Core Xeon 2.66 with 2 GB RAM, 2 x 1.0 Gbps Broadcom NICs teamed for NLB/Fault Tolerance, 2 x RAID 1 - 160 GB Hard Drives.
3 Drives - C:\ - System\OS, D:\ Data, E:\ - External SATA Storage Array (2.8 TB)
There are spread shares on both D:\ and E:\ that users (Windows XP Client workstations) access, about 50 users. The problems is the connectivity is intermittent and it occassionally drops so that when users are saving it takes forever and/or it sometimes "locks" the whole server so UNC browsing and file access is cut off for everyone. Only way to fix has been to reboot.
Contacted Microsoft and they could not find anything wrong. This is, obviously, a big problem as it is a file server. The server is completely updated with all MS patches. An iPerf test from the client workstations to the file server gives an average of 200 MBytes for Transfera nd 180 Mbits/sec.
Event log only shows Errors for the Symantec Tamper Alert (Anti-Virus) due to having the User Hive Profile installed (EID 45).
Any suggestions?
We recently built a new Windows 2003 R2 Standard server on an IBM server. Dual-Core Xeon 2.66 with 2 GB RAM, 2 x 1.0 Gbps Broadcom NICs teamed for NLB/Fault Tolerance, 2 x RAID 1 - 160 GB Hard Drives.
3 Drives - C:\ - System\OS, D:\ Data, E:\ - External SATA Storage Array (2.8 TB)
There are spread shares on both D:\ and E:\ that users (Windows XP Client workstations) access, about 50 users. The problems is the connectivity is intermittent and it occassionally drops so that when users are saving it takes forever and/or it sometimes "locks" the whole server so UNC browsing and file access is cut off for everyone. Only way to fix has been to reboot.
Contacted Microsoft and they could not find anything wrong. This is, obviously, a big problem as it is a file server. The server is completely updated with all MS patches. An iPerf test from the client workstations to the file server gives an average of 200 MBytes for Transfera nd 180 Mbits/sec.
Event log only shows Errors for the Symantec Tamper Alert (Anti-Virus) due to having the User Hive Profile installed (EID 45).
Any suggestions?
ShineOn
Gee, I HOPE you don't have NetBIOS/NetBEUI going on...
ASKER
What do you mean? Everything that was configured was installed as default.
What I mean is this question was cross-posted to the "NetBIOS and NetBEUI" zone and nobody should be using NetBEUI any more - and shouldn't need NetBIOS over IP or WINS either, except for the couple of places they haven't taken that out of Exchange Server yet.
Anyway, I have had issues with Win2K3 R2 dropping connections, and have improved the situation by following a couple of KB docs which I don't happen to have quick-and-easy links to (sorry).
The gist, however, is to change the default autodisconnect timeout on both the server and the client. It involves registry tweaks. If you don't have a problem tweaking your registry on your servers and client PC's, let me know and I'll try to find those KB articles for you.
Depending on your application mix, you may also be experiencing problems caused by opportunistic locking, in conjunction with the auto disconnect. You may want to consider disabling opportunistic locking as well, which also requires registry tweaks.
Anyway, I have had issues with Win2K3 R2 dropping connections, and have improved the situation by following a couple of KB docs which I don't happen to have quick-and-easy links to (sorry).
The gist, however, is to change the default autodisconnect timeout on both the server and the client. It involves registry tweaks. If you don't have a problem tweaking your registry on your servers and client PC's, let me know and I'll try to find those KB articles for you.
Depending on your application mix, you may also be experiencing problems caused by opportunistic locking, in conjunction with the auto disconnect. You may want to consider disabling opportunistic locking as well, which also requires registry tweaks.
ASKER
ShineOn,
I beleved it may had been related to NetBIOS.. or more correctly WINS as when the server was first setup and the NIC was assigned a static IP, it was not able to be hit by UNC name without the WINS server setting hard coded to the NIC.
If you can send me the links/docs for the autodisconnect/opportunist
I beleved it may had been related to NetBIOS.. or more correctly WINS as when the server was first setup and the NIC was assigned a static IP, it was not able to be hit by UNC name without the WINS server setting hard coded to the NIC.
If you can send me the links/docs for the autodisconnect/opportunist
Disable auto disconnect: http://support.microsoft.com/default.aspx?scid=kb;en-us;297684
Note that they confuse the issue by throwing the KeepConn setting for the client in the middle of the registry change for the server side... The autodisconnect entry goes in the server's LANMANSERVER key and the keepconn setting goes in the workstations LANMANWORKSTATION key.
Configuring Opportunistic Locking: http://support.microsoft.com/kb/296264
Note that the workstation setting is opposite that of the server - the server is "EnableOplocks" and the workstation is "OplocksDisabled" so the workstation's setting should be 1 and the server's should be 0 to disable both requesting and allowing of opportunistic locking.
Hope this helps.
As to the other thing - you shouldn't need WINS and NetBT if you have your DNS configured right... maybe a topic for another Question?
Note that they confuse the issue by throwing the KeepConn setting for the client in the middle of the registry change for the server side... The autodisconnect entry goes in the server's LANMANSERVER key and the keepconn setting goes in the workstations LANMANWORKSTATION key.
Configuring Opportunistic Locking: http://support.microsoft.com/kb/296264
Note that the workstation setting is opposite that of the server - the server is "EnableOplocks" and the workstation is "OplocksDisabled" so the workstation's setting should be 1 and the server's should be 0 to disable both requesting and allowing of opportunistic locking.
Hope this helps.
As to the other thing - you shouldn't need WINS and NetBT if you have your DNS configured right... maybe a topic for another Question?
Note that you MAY see a decrease in performance opening files across the LAN, by disabling oplocks. Try the disconnect thing first and see how things go.
If the problems don't quite go away after changing the disconnect, but do after disabling oplocks, it may be some 3rd-party tools you might have installed, like an open-file manager for a backup system, or an antivirus/firewall product masking the "break oplock" event causing it to be mishandled at the server.
If the problems don't quite go away after changing the disconnect, but do after disabling oplocks, it may be some 3rd-party tools you might have installed, like an open-file manager for a backup system, or an antivirus/firewall product masking the "break oplock" event causing it to be mishandled at the server.
ASKER
Yes, there is the BrightStor ARCserve Open File Agent (11.5) installed/configured.
As far as AV it is Symantec Anti-Virus 10.x
As far as AV it is Symantec Anti-Virus 10.x
See this KB article: http://support.microsoft.com/kb/822219
ASKER
Thanks for the find.
But still does it not sound odd that when the file server "drops" connections it then freezes all network user workstations so they can't even navigate, open apps, even task manager?
But still does it not sound odd that when the file server "drops" connections it then freezes all network user workstations so they can't even navigate, open apps, even task manager?
Never had that happen with NetWare on the back-end, but everyone seems to think Windows is a superior solution. Go figure.
It's odd, but my gut says it's got something to do with either the timeout or the timeout in conjunction with oplocks.
It's odd, but my gut says it's got something to do with either the timeout or the timeout in conjunction with oplocks.
ASKER
Symantec Anti-Virus seems to have been the problem.
The AV engine was set to lock network connections rather than log them in regards to tamper alert protection.
The AV engine was set to lock network connections rather than log them in regards to tamper alert protection.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Does anyone know where the SAV setting is to lower the sensitivity?
On other servers it seems to have the same problem with Tamper Protection blocking the User Hive Profile cleanup.
On other servers it seems to have the same problem with Tamper Protection blocking the User Hive Profile cleanup.