Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us $3,400 per month for bandwidth bills alone, and since we don't believe in shoving popup ads to our registered users, we try to make the money back through forum registrations.
  • Post
  • Reply
madsushi
Apr 19, 2009

Baller.

The Fool posted:

So, after doing some stuff with eseutil, I am getting an error in event viewer that says the db is not mounting because the site and org name are different. After doing some research it appears the only way to resolve this without any downtime on the customers part is going to be:

1. Setup a test install in vmware at my shop, matching the original site and org names
2. Restore db's to test environment.
3. use either exmerge or an outlook client to export what I need to pst
4. import from pst to new server

Am I right in thinking that this is probably the best way to do this?

Yes. I have had to do this before, and this was the only way I found to fix it.

Adbot
ADBOT LOVES YOU

madsushi
Apr 19, 2009

Baller.

I had a similar issue where RPC users had to be in the office before their account would work. It was due to an Exchange/Domain Controller issue, where the Exchange proxy service wasn't able to talk to the domain controller to find their user name. When you're local, your PC just goes to the DC directly. I had to add a whole bunch of registry entries to get that communication working again.

madsushi
Apr 19, 2009

Baller.

I have seen this issue pop up on Exchange 2010, and I am wondering if any of you have seen it or know of a fix.

A user sends an email to say 10 external recipients. 9 of those 10 recipients are valid, but 1 of them isn't. The email goes out to the 9 mailboxes, but the 10th one gets rejected (with say a 4xx or 5xx error) and the entire message goes back in the Exchange queue. A minute later, the email goes out again to the 9 mailboxes, but the 10th one gets rejected. Repeat indefinitely until the 9 working users all get TONS of copies of the email, while the messages is still stuck in my Exchange queue waiting on the 10th person's mailbox to start working.

It seems like Exchange doesn't split the message into 10 different pieces, but rather sends it as one big push and will restart from scratch any time one piece of the push fails.

madsushi
Apr 19, 2009

Baller.

Linux Nazi posted:

This isn't typical. The bad recipient should generate an NDR that gets kicked back to the sender and that is it. I've not seen it behave any other way. Even if it is sent to a distro group with external contacts as members, and one of those contacts is bad it should gracefully kick back an NDR to the sender.

I can't imagine a scenario where this would not be the case, unless you are relaying to a smarthost that is re-submitting the messages.

We are relaying through a smarthost outbound (Postini), but the message definitely gets stuck in our queue and keeps retrying. If the error code we get back is serious (e.g. no such mailbox) it works fine, but if the error code is like "Please Retry Later" then it just keeps trying. I will see if it occurs if we don't relay through Postini. Thanks for the tip.

madsushi
Apr 19, 2009

Baller.

Actually, I recommend keeping your 2010 stores/databases as large as possible. Every new store you introduce also introduces another maintenance thread, which consumes IOPS constantly throughout the day.

I read a great article from NetApp about their SAN testing with Exchange 2010. They had to host something like 20 TB of mail and get a certain level of performance/latency out of it to be certified by Microsoft. When they were using all 500 GB databases (so 40 in total), they couldn't meet Microsoft's requirements for latency, because of all of the maintenance threads consuming IOPS. When they consolidated databases and shrunk down to 10 databases of 2 TB each, they passed the test with flying colors.

Exchange 2010 is designed to constantly utilize your disks for maintenance rather than trying to cram a LOT of IOPS into a maintenance window. Every database you add means that another maintenance thread starts as well.

madsushi
Apr 19, 2009

Baller.

Spamtron7000 posted:

Thanks. That sounds great. I can see how it sounds like a strange question so I'll describe my requirements. We run a hosted Exchange environment for many different small companies who are affiliated with the parent company (my company) but not affiliated with one another. We have an SPLA agreement with Microsoft to support this. I really hate that the members of the small companies can see each other - in fact sometimes it's bad for business because they like to get together and bash us behind our backs

Actually, GAL separation is in Exch2010 SP2.

http://blogs.technet.com/b/exchange...27/3411882.aspx

http://blogs.technet.com/b/manjubn/...-practices.aspx

madsushi
Apr 19, 2009

Baller.

Be careful with doing a clustered CAS/HT front-end server if you're still running Outlook 2003, I ran into an issue with it that required Outlook 2003 SP3 which wasn't compatible with some business apps and so we had to kill the cluster.

madsushi
Apr 19, 2009

Baller.

Linux Nazi posted:

IIRC there's a problem with public folder connections and static RPC ports, but if you are moving to a DAG you can no longer utilize public folders anyways. This is also why you do not have to worry about Outlook 2003, it requires public folders to operate, and public folders cannot be made highly available, or be included in a DAG. So you will be removing any Outlook 2003 clients prior to building it out.

Honestly the way you are going about it is actually less simple, building out 2 VMs and installing Exchange with the required roles would take maybe 5 hours total, and can be accomplished in a production environment with no impact until you've got everything configured and are ready to make the cut-over. It certainly won't cost as much as a hardware balancer.

Remember, you get (with some restrictions) 4 VM installs per Server 2008 Enterprise license. So you may have some licenses available for the guest VMs already.

You can have Public Folders and have a DAG, your PFs just won't fail over (which may or may not be acceptable).

madsushi
Apr 19, 2009

Baller.

This is probably a good thread for this question:

Mail filtering.

Currently we use Postini, but there's a rumor going around that they might be going the way of the dodo. We're trying to find a good option that provides inbound/outbound filtering, is located in the cloud, and has perks (like outbound encryption, etc). A major required feature would be the ability to host multiple domains/customers underneath one account. We have heard of a good solution, MX Logic, but I wanted to see if anyone has any good/bad experiences with MX Logic or if there's another service they'd recommend.

Thanks!

madsushi
Apr 19, 2009

Baller.

Powdered Toast Man posted:

I loving HATE NETAPP SNAPMANAGER FOR EXCHANGE.

That is all. Thank you.



Why is that? I have found doing SMBR restores to be the best thing since sliced bread. You just have to have your LUN and DB/Log layout set up and you're good to go.

madsushi
Apr 19, 2009

Baller.

Frozen-Solid posted:

With a server like Exchange, how would you determine if you need more memory or not? I know the database will do it's best to cache as much as it can, but does it ever reach a point where it doesn't feel like it has to keep sucking up memory for caching?

Right now we are floating between 8.3-8.5 GB out of 10 GB free, and it seems to be running just great. We have some spare memory that I can allocate to it, but I'm not sure if it's worth the increase or not. If I give it another 2-4 GB, would it just start using the same % of memory, or should I only worry about allocating more if it gets to the point it's using close to 100% of it's memory?

It will chew up a lot of memory, but that's normal. You can use the Microsoft sizing guides to figure out how much memory your environment should need. How many users do you have, and what's your mailbox count (and total DB size)?

madsushi
Apr 19, 2009

Baller.

Frozen-Solid posted:

We're relatively tiny. 75 mailboxes, total mailbox DB size is 25 GB. The minimum for a single server with all 4 roles was 10 GB, which is what I gave the VM to start with. My original sizing estimates based on Microsoft's guides were 8-12 GB. We're not using Unified Messaging, so 2 GB for each role + 2 GB cache made 8 GB. The consulting company we hire insisted we absolutely couldn't make an Exchange server without 16 GB of memory which seems ridiculous to me.

Since we're at 8.3/10GB right now, I'm not sure what actual % I should expect, or how I would tell if I SHOULD give it another few gigs. I obviously don't want to over allocate, since it's a VM and if it doesn't need it, it's not going to get it.

10 GB is more than plenty for your environment. I have one client environment with 150 mailboxes and about a 100 GB database running on a single 8 GB server without issue. I would keep an eye on the event logs but wouldn't concern myself with that at all.

madsushi
Apr 19, 2009

Baller.

Syano posted:

Keep in mind 2 server dag only gives you ha of your info stores. If one server of the two dies no one is still going to get email because you haven't got ha of your cas role. Only supported way to get ha of your cas role is thru a cas array which uses NLb which does not Coexist with a dag cluster. You can futz around it with some ad trickery but just know what you're doing before you try

I didn't think you could even do a DAG at all with the CAS/HT roles installed on the MB servers. I thought both MB servers had to be pure for that to work?

Adbot
ADBOT LOVES YOU

madsushi
Apr 19, 2009

Baller.

MigrationWiz is baller as hell. Used it like 40-50 times to move people to my last company's hosted Exchange environment, off of a ton of random services, from local Exchange to 365, from 365 to our hosted Exchange, etc. Works like a charm, easy pricing, etc. Highly recommend.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply