[Host Unreachable] Exchange 5.5 Internet mail service 5

[gp8]

Hi , 99% of mails leave the IMS OK but a few will just sit and when looking at the details it says [Host unreachable]. I have tried nslookup to the host and get a valid response, I have even managed to ping the host in some instances
I have tried different dns servers address's with no result.
I have set the dns cache on the exchange server to force a dns resolution every time it sends a mail.
Any Ideas anyone???

Ged

 

[tcswan]

I can pretty much rule out the user config side.. Already went down that road in the very beginning.. At least in my case.. tried sending to several of the problem domains from multiple clients with the same results on each client. Messages are getting from the client to the IMS queue, but stopping there with a host unreachable error.. Definitely not a user config issue.

 

[msdIT]

Kilfarsnar
When I switched on the Exchange Event logs on, It just created entries in the NT Event Log. It didn't report anything that I didn't already know.

The error message that I wrote in my last reply came from the Exchanges /IMCData/Log/.

Unsure if that is what you mean.

Has anybody set up a Test Server and Installed a fresh copy of NT and Exchange? I suspect it relates to a NT registry entry. This would explain why a reinstall of Exchange alone did not help those that tried it.

Also, this seems a common issue. Has ANYBODY called MS about it?

 

[kilfarsnar]

msdIT,
When I did my rebuild a few weeks ago I installed a fresh copy of NT4 SP6a and Exchange 5.5 on a completely different set of hardware. As I said, the only thing the two installs have in common is the DB's. It has not resolved the issue. What I was referring to is the settings in the properties of the IMS conneter. There is a tab called Diagnostics logging. Here you can set logging detail levels for a number of aspects of the IMC. Unfortunately, I have nothing in the exchsrvr\imcdata\log directory but the route.txt file. Stupid question, but does where is the SMTP protocol log is located? In light of recent posts, I'd like to look at it. I am also going to see if my boss will approve me calling M$ today.

 

[msdIT]

Kilfarsnar.
I can only say, Mmmmmmmmm, interesting that you did a fresh install and still have the same issue. This has me thinking along a different line now.

Several of my clients running simular if not identical types of Servers and software. Only ONE has the problem.

I'll let you know the settings of this when I am there next. I have looked and spoken to experts on this and to date am no better off. I (We)(All) are going to find the solution to this as it seems to be a BIG headache.

Anybody want a challange on a race to the solution.


Lets get this resolved once and for all and end this thread!

Kilfarsnar. I'll get back to you on your queries

msdIT

 

[kilfarsnar]

Heh, and I just re-read one of my previous posts. What I meant to ask is does anyone know where the SMTP protocol log is located? Sheesh, I gotta read before I click!

 

[msdIT]

kilfarsnar;

This particular issue has me very interested but yet without an answer. I have done interesting and some very intensive tests with the help of some Gurus and "Host Unreachabe" still STANDS OUT and unresolved. I have learnt a lot in the process "all for free"

SMTP Protocol Log: So you think you may be onto something...well, Have fun.

PS: I doubt Microsoft will give a direct answer but who knows. Maybe the reason we haven't heard from anybody who HAS called MS is that they have resolved the issue hence have no need to return to this Thread?

Happy Hunting

 

[Fairbankstech]

I don't have any new information to add, but I'd like to chime in about the location of the SMTP logs.

As said before, you need to enable Diagnostic Logging in the IMC section of the Exchange Administrator.

I have the following options set to minimal: Message Transfer, SMTP Interface Events, Internal Processing, and SMTP Protocol Log. It appears that the key item though is the SMTP Protocol Log.

You can determine which log the a Host Unreachable failure was written to by looking for the "!" messages in the Application section of the Event Log. It will reference the correct log number, then you just match the name or time.

The logs should be located in /exchsrvr/imcdata/log/. This folder could be on different hard drive than expected if you have recently run the Exchange optimizer or recreated your IMC (which is what happened to me.)

Hope this helps!
BJ

 

[tcswan]

Kilfarsnar:
The fact that you did a new install from scratch on different hardware with everything fresh except the DB's, takes me back to something breaking down in the receiving domains interpreting of some piece of information. Something different about the receiving domain's configuration, and how it's interpreting a piece of header information or something. I ran across the below pasted notes referencing msdIT's log error and the 501 Syntactically invalid HELO argument(s). This is not the problem in my case, but it may point to something else in the right general direction.

--------------Paste---------------

Syntactically invalid HELO/EHLO.
Description
The HELO or EHLO command is followed by the remote server's hostname. The domain passed with the command should be a valid one. This problem is most noticeable on Windows servers where they often configured with the same DNS name as their local 'windows' domain, which allow '_' to be contained within the domain. When the remote server issues an HELO or EHLO, it includes that same local windows domain name, rather than a syntactically correct, valid, hostname.

Note that while SMTP servers should not block mail with an invalid hostname passed to HELO or EHLO, many do. Although it is not technically breaking RFC's, the fact is that the [Sending] mail server is incorrectly configured.

----------End-Paste---------------

We have no underscores in either case, so this is not the exact issue in our case, but it makes me wonder if there's not something else that IS configured in such a way that the HELO is being rejected. I've verified our RDNS, so that's not it, I've verified we're not blacklisted by either the local receiving server, nor any of the blacklisting services. I can telnet into the mailservers I tested so I have to wonder if it's not something in the header information being stamped on the message. You guys actually seem much more knowledgeable about Exchange than I am so maybe this information will spark a thought on your end.
Question: in looking at the outbound log files on these messages, the message ID reads <message ID#>@<computerame>.
Computername is a valid computer name, but could the receiving side be looking for a differently formatted name (IE: a domain name, or comparing this name to a RDNS or something here?). Some of this may seem incredibly elemental to you guys, but I sort of inherited this in a one man IT department, so bare with me here..

 

[kilfarsnar]

tcswan,

Yes, I agree. I have often thought that this could be a problem on the receiving end. I haven't taken many steps to investigate this at this point. What is really maddening to me is that all of my problems are internal! My domain does not have any dashes or underscores. I am sending mail from one exchange server to another within the same logical organization. I am specifying that server by raw IP address. I can ping, telnet, browse the file system, and even send mail while this failure is happening! I am on with M$ right now. I feel for the guy on the other end of the line. I'll keep y'all posted.

 

[kilfarsnar]

Ok, I am off with M$. I have updated the versions of the IMC and MTA executables. The donload links are:

http://www.microsoft.com/downloads/...9C-611B-4E1F-95C3-9FC99A0511A3&displaylang=en

and

http://www.microsoft.com/downloads/...82-c42a-4f27-adc0-27dd3874d8d9&displaylang=en

I will be back in touch with them next week to check status. So apply the patches and see how it goes. I am keeping my fingers crossed.

 

[kilfarsnar]

Hmm, it looks like the second link got messed up. The second link is:

http://www.microsoft.com/downloads/...82-c42a-4f27-adc0-27dd3874d8d9&displaylang=en

 

[msdIT]

kilfarsnar,
Did the updates work for you?

 

[tcswan]

I don't know.. Not willing to rule anything out yet, but it kind of looks like M$ was shooting in the dark here a bit.. The first update I could see as a maybe, don't really see the 2nd update applying to this issue, but please let us know..

 

[Fairbankstech]

I agree with tcswan... I've had these updates installed since they were released. (Check your add/remove programs panel)

 

[msdIT]

Thanks,
I am never keen on installing service packs that don't relate to a specific issue. In this case neither look as though they do. I read up on both and feel you are right to say M$ are shooting in the dark. That's not unusual, it is luck of the draw what Tech Person you get on the other end. In the past I have spoken to a few "No Idea Techs" and some really great ones.

Anyway, I won't rule it out but will wait to hear if it worked for you. Can't you do a test or something?

 

[kilfarsnar]

No, the updates did not work. Had a failure over the weekend. And yes, M$ is definitely shooting in the dark. I can't say I blame them. This problem makes little sense to me, and I've been thinking about it since last June! That's why I said I felt for the tech I spoke with. I could almost hear him scratching his head. Next they want copies of my firewall logs, which I have to request from my monitoring company. Because of the nature of my issue, I will also investigate sending only replication messages over my WAN link, and having the rest go over the gateway. I haven't had any issues with mail that goes over the gateway. We'll see...

 

[msdIT]

Agreed, it is a real headache this error.

I am mentioning the below as a revision of the problem. I don't believe Firewalls are a problem here due to my testing. But I'm not the expert.

My Scanario:
All tests are done on the Exchange 5.5 Server (NT4 SP6).

Test 1: Email will send without issue if I set Outlook (or Express) to send directly via the ISP.
Test 2: The same email will not go if I set it to send though Exchange Server. "Exchange is set to use the Servers DNS settings".
Test 3: From the command prompt, I can Ping - Trace - Telnet to the offending email address without issue.

I am surprised that M$ have a problem with assisting here as it seems a real issue for a lot of people. Maybe you should send them to this thread. There are many more like it. Keep on trying. May even be worth checking out other news groups.

 

[tcswan]

I agree w/ you re: not blaming M$ for shooting in the dark.. We've all been doing that for months.. Sooner or later they'll end up having to escalate it to higher levels as long as kilfarsnar keeps bugging them about it, which, since I'm sure they've already charged him for the incident, he will. Got to thank all you guys though.. At least I feel like I'm once again heading towards a solution to this for the first time in months.

 

[gp8]

Hello All,

I agree totally with msdIT regarding installing MS patches, I have lost servers in the past causing me a lot of grief. The use of the "_" in the host name is something that has been in the back of my mind for a while. I have discussed this with my ISP and another supplier who both agree this could be one cause of the problem. I now intend to setup an SMTP relay server (without the "_" in the hostname) out on a DMZ and have my mx records changed to suit. This will need to be done out of hours because of the time it takes for the new dns to propagate it will not be until this weekend, I'll let you know how I get on.

cheers

 

[tcswan]

GP8.. The use of an underscore in your case may either be the problem, or be contributing to the problem in your case. There's still another issue though regarding this though because several of us experiencing this problem do not have underscores in our configuration.

Good luck, and let us know if this resolves the problem for you..