| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 674.1 | Still slow | VAXRIO::DJOPC4::SCHOR | Email: VAXRIO::SCHOR | Tue Mar 18 1997 09:07 | 26 |
| Hi,
We had some network problems.
We solved them but the login is still slow.
Wfwg client logon now lasts 20 seconds until the begin of the execution of
the login script (with Cluster NT client installed).
If we desinstall the cluster client software it only lasts 5 seconds.
Is this a normal behaviour ?
I collected an IRIS trace and saw that Cluster client is trying to resolve
an NetBIOS name (node<DC>) for 14 seconds ...
There's no response from broadcast neither from the wins server. Even if I
put this name is lmhosts, the client software doesn't send any message
during 14 seconds !!
Is this normal ?
W95 doesn't try to resolve this name.
Any ideas ?
Thanks,
Helio.
|
| 674.2 | | LJSRV1::GOODMAN | | Wed Mar 19 1997 10:05 | 2 |
| Exactly what is the login script? Are the WFW machines logging in after one
of the cluster nodes is shut off?
|
| 674.3 | asnwers | VAXRIO::DJOPC4::SCHOR | Email: VAXRIO::SCHOR | Wed Mar 19 1997 15:35 | 21 |
| Hi,
> Exactly what is the login script? Are the WFW machines logging in after one
> of the cluster nodes is shut off?
The login script has some net use connections to the cluster file shares.
net use s: \\clstralias\scratch /y
net use g: /home /y
net use u: \\clstralias\org /y
net use p: \\clstralias\producao /y
net time \\node1 /set
net time \\node1 /set
I can see in IRIS trace that cluster client software search for the
NETBIOS name before reading the script file.
Thanks
Helio.
|
| 674.4 | exceprts from the documentation | CSC32::V_STROUD | | Wed Mar 19 1997 16:17 | 31 |
|
On page 5-11 in the Digital Clusters for Windows NT Configuration and
Installation Guide of the new documentation for 1.1 and SP2, it talks about
adjusting the time for trying network drive connection for WFW clients. It
then refers you to the Appendix A of the Digital Clusters for Windows NT
Administrator's Guide page A-25.
It talks about request for network connects, such as network drives can
take up
to 15 seconds to fail. You can adjust this time period by changing the settings
in the SYSTEM.INI file.
To change the timeout period or retry count for a Windows for Workgroups client:
1. Add the following section to the system.ini file:
[DCLNAMC]
TimeoutPeriod=5
RetryCount=3
2. Set the TimeoutPeriod and RetryCount to the desired values.
3. Reboot the system.
A colleague of mine put the timeout to 3 seconds and the retry count to 1 and
it had a significant improvement in the performance for the WFW clients.
|
| 674.5 | Goog workaround | VAXRIO::DJOPC4::SCHOR | Email: VAXRIO::SCHOR | Thu Mar 20 1997 15:36 | 18 |
| Hi,
This is a very good work around !
I'd like to know:
1- Is the normal behaviour ?
2- Is there any reason for searching this netbios name ?
The node is the same node that validates the PC logon.
Win95 clients don't look for this netbios name.
Thanks,
Helio.
|
| 674.6 | The reason .... | VAXRIO::DJOPC4::SCHOR | Email: VAXRIO::SCHOR | Fri Mar 21 1997 07:16 | 17 |
| Hi,
In the beginning of the PC logon process, it connects to \\dmncontrol\IPC$.
Where dmncontrol is the domain controller (PDC or BDC) that validates
the PC logon.
If it's a W95 client, the cluster client software tries to resolve
NETBIOS name dmncontrol<0x20>.
If it's a wfw client, the cluster client software tries to resolve
NETBIOS name dmncontrol<0xDC>. The cluster name is not dmncontrol.
After 15 seconds (timeout) it triies dmncontrol<0x20>.
Is there any reason for this different behavior ?
Thanks,
Helio
|