MS Outlook 2003/2007/Exchange 2003 requires ipconfig /flushdns to send emails

is there a pattern to where the effected users are located?

and have you checked the dns settings for the effected machines?

you have a problem with your dns and it is providing your clients with the wrong information

I'd also check your wins servers and see if you don't have a rogue server on the network

in fact - the next time one of your user's reports a problem, have them a) ping the exchange box to see where their machine thinks it is, and then give you the dump of an ipconfig /all to see where they are pulling their information from

mvgeertruyen:

   Yes it is clearly dns related, hence the /flushdns to clear the cache and "fix" the issue.  Yes, the mail server is "available" to the internet.  Different names for internal vs external users.  External would be sending mail to our mx record mail.domainname.com which actually points to a IronMail Spam Filter appliance and then from there it goes to our Exchange server.  External mails coming in have no issues.  Internal mails going out have issues.  As I have said, a few of the users have to /flushdns to get their mail to go while others are getting an erroneous DNS error saying that the domain in question does not exist.  And these are well known domains like qwest.com or verizon.com, etc, etc.


abhaigh:

    No pattern at all, it's totally random and is only currently affecting 3 people in the corporate office and 2 at remote sites.
    All machines are DHCP'd and get their DNS from our primary name server, which everyone uses.
    How does one check to see if there are extra wins server on the network?
    This is currently ongoing andf I did ping the mail server from the clients machine and it's pointing to the correct address.  /all chows them getting their info from the configured DHCP server and all properties are correct, DNS, gateway, etc, etc.

I don't know your environment but this comes to mind:

- Perhaps your dns queries are timing out - you can test this with nslookup set debug. If this is the case then the client will mark it as unreachable and will not retry until it's cache expires (what you do with flushdns). I you suspect anything along these lines you can set the query timeout on the dns server.

- Might be worth checking the cache of your dns servers for invalid ip's (don't think it is this as the clients would resolve to invalid address)

run an av scan on these machines - I'm thinking that the problems are local to these machines as if it were central then more machines would be effected

We run McAfee Anti-Virus Enterprise, all machines are current and all machines have latest definitions.  Verified by SCCM2007 and verified by running local scans when first reported.

If it is network related then the first thing that always needs to be checked is check the network card drivers.

Simon.

All network card drives have been verified as being the latest updates from Dell themselves.

check for spyware too

Guys, all of the standard checks and scans have been completed.  I've run anti-spyware, anti-virus, anti-malware and anti-wareware.  All system drivers have been updated and verified.  All system level patches released by MS have been installed.  We're running Windows XP SP3 on the clients.

What we have here is something beyond the mundane.

Latest network card drivers from Dell does not mean they are the latest drivers available. I have lost count the number of times I have had to get drivers from the hardware supplier directly. For example, with your server, unless the network card driver is dated 2008 or later, then it is not up to date.

Simon.

Are you seeing any errors on the dns servers? Can yuo increase logging on them? How many dns servers do you have? Are all clients using the same or different ones? Forwarders?