TechNet Top Questions - February 2001

Article
12/22/2008

What Time Is It In Exchange 5.5?

I have exchange 5.5 with sp4.When anyone sent an email, the receiver receives it after about 4 hours plus the time which is on the email is way out, like 4 or 5 am.

Does anyone know how to set or reset the timer on the exchange?

Thank you.

A: "Eric" (eap.neinspam@spamless.bratrud.com) replied with:

Make sure that the Time Zone on the server corresponds to your local time zone. Windows NT time codes everything based on GMT, but displays the time as an offset of GMT based on the local time zone. IOW, if I am in NY and my server's time zone is set for GMT, the messages people in NY receive from my server will be offset by 5 hours. BTW, if you are not paying attention when you set up an NT4 server, the Time Zone will be GMT by default.

Also check out 262802 - XADM: Server Monitor Clock Synchronize Option Sets the Incorrect Time on an Exchange Server-Based Computer

IIS Compression

Q: "Ilia Shapira" < ilia@imvamp.com> wrote :

I have a very interesting question about the subj. I'm using HTTP compression on my IIS server but I noticed that when users don't write the full web address, IIS doesn't compress the files.

For example:

My site is www.test.com.

My IIS default document is index.html.

When a user writes www.test.com he gets index.html (because its configured as default document in IIS) but IIS doesn't compress this file!

When the user writes www.test.com/index.html he gets the same file but this time IIS compresses this file!

Is there a way to do something with this problem?

10x

A: Lets look at what is really happening with this. When a file is requested, the compressed copy is stored in a temporary folder, and then sent when further requests are received. Files are not compressed prior to receiving a request for them.

255844 - Default Document Is Not Compressed When HTTP Compression Is Enabled

SMS Installer to Re-Package?

Q: "jamal" < indiana@yahoo.com> wrote:

I am trying to automatically install Adobe reader. I have created the package and advertised the program. It successfully sends the package out to the clients and comes up with the installer window.

But what I want to know is if there is anyway of installing the Adobe reader without having to click next/without any input from the administrator.

Many thanks

A: "Jeff Hartley" < jeff.w.hartley@mindspring.com>:

Use SMS Installer to repackage Acrobat. When the repackage is complete, go in and remove all the installation dialog boxes. There is an easy

GUI to assist you with this so you don't even have to edit script.

If you haven't repackaged Acrobat, then you need to install sms installer from the SMS cd and learn how to do a repackage. I can't explain all of it here.

There are a few good books out there and you can get example scripts and tips at https://www.swynk.com/trent.

If you have repackaged Acrobat using sms installer then all you need to know is how to get rid of all the dialog boxes:

Open the acrobat .ipf script with sms installer.
Click 'installation expert' on the 'view' menu
On the left side, double-click 'installation interface'
Click the 'dialogs' tab
Remove all the check marks
Re-compile.

In reply to this, "sms admin" < <lkj;lkj@yuojh.com>> sent:

Actually Adobe comes built in with an automated install feature. Do a setup.exe -r to record the settings. Then a setup.iss file will appear in the Windows directory of the pc you ran the install on. Take that .iss file, put it in the installation source files, and run setup.exe -s to run it silently.

Generally you shouldn't use sms installer unless the program you're trying to install doesn't already have a built-in install automation process.

And we concur. The Rule of Thumb is: If the Software comes with an unattended install method, use it.

Also see 180918 - SMSINST: Guidelines for Using Systems Management Server Installer

SQL Server CPU Peaking

Q: "WB" < wb@msn.com> wrote:

What would be the events, filters etc to specify in a SQL Profiler trace to determine what processes are peaking the CPU for more than 3 or 4 seconds.

In Perfmon there are occasionaly spikes above 90 which last over 5 seconds, and I'd like to find out exactly what's doing that via a trace. However, in Profiler, the CPU column (and filter) seems to only specify CPU time for a specific event, not the duration of peak CPU time.

Thanks

A: "Andrew J. Kelly" < akelly@targitmail.com> replied:

It's normal to have spikes in the CPU. Something as simple as a select with an ORDER BY clause can cause a large CPU spike but is perfectly normal. You can run a general trace in profiler and put the results to a table. Then just select against that table to see the higher cpu bound tasks and that should point you in the right direction.

Andrew J. Kelly

Windows 2000 Symmetric Clustering?

Q: Dave" < dlmo@dynegy.com> sent:

I have set up a Windows 2000 symmetric cluster but am having some difficulties. Since I have 2 servers each owning a different file share, my intent was to have the cluster server showing both shares when accessed. For example, I have Server1a that owns SVR1aShare and Server1b that owns SVR1bShare. If you accessed Server1 (name of the cluster server) I wanted to be able to see both shares, SVR1aShare and SVR1bShare. It makes a central point to map drives to. If one of my servers fails then it will fail over to the other one and still both shares will be visible through Server1.

I cannot seem to get this to work. Whoever owns the Cluster Group (Server1a for example) shows its ownership when you access Server1 (i.e., when Server1a owns the Cluster Group and SVR1aShare (SVR1bShare is still owned by Server1b) and you access Server1, you only see SVR1bShare, not SVR1bShare.) I got around it, sorta, by making new IP address resources and Network Name resources that have both servers as possible owners. but then you still have mappings to the network name (which is different for both but can failover ownership) not the cluster name, Server1.

Almost all other NOS's can do this.

A: "David Devlin" < devlin@canada.com> replied:

Dave,

Microsoft uses a share-nothing model. This means that clustered resources fail over as a complete group. All the resources within the group must stay together and can only be owned by one server at a time.

Dependencies play a big part in why this is the way it is. For example: the NETWORK NAME resource is dependent on the IP ADDRESS resource, or, the FILE SHARE resource is dependent on the PHYSICAL DISK resource. That means if any part of the resource group fails over the whole group has to fail with it.

To do what you want to do you have to create a NETWORK NAME resource, IP ADDRESS resource, PHYSICAL DISK resource (or 2) and 2 FILE SHARE resources. Then you can map to the NETWORK name resource and both FILE SHARE resources will be available. Unfortunately this means that one node will do all the processing for this clustered group, you cannot "share" processing, - it's all or nothing. However, you can create as many independent groups as you like...

What we did at our site was this:

We keep the original clustered resource for administration only (owner server1).

We created a FILE SHARE resource group for shares (preferred owner server2).

We created a MSSQL resource group for SQL server (preferred owner server1).

This helped split the processing between the nodes but gave us the redundancy and

Fail over we wanted.

Hope this helps?

David Devlin

Great help, David! We would just add that the Cluster Virtual Name (Cluster Name) should be used only for administration purposes.

Name abstraction is a central concept in clustering. To understand name abstraction, consider that if you installed Microsoft Exchange Server in the past, you inherited a namespace from that installation – in this case, the host name. So every time you refer to the Exchange server, you have a dependency host name, and every client that gets to that Exchange server gets there by using the host name.

Within clustering, you're taking that namespace and removing the physical link and physical dependency of where the name actually exists. This means that the name no longer exists within the context of a specific node, but instead is "floated" as necessary. The same is true for other types of services and resource. With name abstraction, you can also take an application or service and reinstate it where you need it to run. With nothing more than a service name, clients can use the service regardless of what happens to be running at any time.

No Ping Through Proxy

Q: "Steve" < stefan.kainz@freshfieldsbruckhaus.com> wrote:

Dear All!

I just don't get it. I cannot ping from any workstation to the Internet. I use a MSProxy 2.0 and a firewall. I can ping from the Proxy itself, no probs.

Why does the Proxy block ICMP echo??

On default, ICMP is open on the proxy and I did not change that...

On clue more, I have IP-forwarding enabled on my internal network card (I have a reason for that...), and I think since then the ping does not work anymore...

I would appreciate any help!!

Kind Regards

Steve

A: "Andrew" < itsec_andrew@hotmail.com> replies:

You will never be able to ping through Proxy. It just doesn't support it. Nothing you do on the firewall part helps.

Technically, the reason is that ping uses ICMP, and not TCP/UDP. The winsock proxy client on the workstations only deals with winsock applications – i.e., TCP or UDP.

Sorry I can't give you a magic solution :(

Andrew

Ping will resolve the name however, but will not receive the return packets.