Posted by: Anthony Drewery | July 31, 2006

Resetting the MIMEsweeper SpamLogic database

We run Clearswift’s MIMEsweeper for SMTP 5.2 as our SMTP gateway. We have 4 policy servers (2 x Tampa, 2 x Crawley) connected to one configuration server (Crawley). I’ll spare you my rant about Clearswift and MIMEsweeeper but suffice to say I won’t be recommending it to anyone who needs a reliable messaging hygiene solution for a high volume enterprise environment. (FWIW, I’ve been working with MIMEsweeper since 1998 and until its current incarnation have always been a stong supporter of the product.)

We’ve had a plethora of issues with MIMEsweeper version 5 but one in particular keeps re-occurring. Despite having applied their hotfix, every few weeks MIMEsweeper’s spam solution, SpamLogic, takes on a mind of it’s own and starts trapping an unacceptable amount of false positives. When this happens the SpamLogic databases need to be reset to their default configuration. To do this connect to each policy server in turn and do the following:

  • Go to the c:\Program Files\Clearswift\MIMESweeper for SMTP\Data\SpamLogic folder.
  • Rename spamlogic.db to spamlogic.old.
  • Copy spamlogic.bak to spamlogic.db.

This way the original spamlogic.bak is preserved as you’re going to be needing it again in the future! Once you’ve completed these steps restart the MIMEsweeper Security Service.

You can disable the SpamLogic auto-training feature at anytime but then your databases will remain static and your detection rates drop as the type of spam evolves.

[tags]MIMEsweeper[/tags] 

Posted by: Anthony Drewery | May 16, 2006

Battlefield lessons

We have a total of 4 Storage Area Networks attached to our primary Exchange clusters. We have 2 HP MSA 1000′s , an EMC CLARiiON CX300 and an EMC CLARiiON CX500. When designing our storage solutions I try to ensure a maximum database size of 50GB and leave twice that amount of free space so there is room for DB maintenance if necessary. I would normally put all databases for a storage group on a single RAID 1+0 array. This way there’s a little give and take with the DB sizes. For example if I planned to have 4 mailbox stores in a storage group I would make the partition at least 300GB (4 x 50Gb for stores + 2 x 50GB free space for maintenance).

Unfortunately I haven’t been as strict with our older hardware and have learnt an important "battlefield lesson" (thanks go to Gary for the post title). We have a server (an old HP ML570 ) that we have been using as a transitional area for mailboxes that we’re manually archiving. As it doesn’t normally contain much production data we’ve been a little lax with our housekeeping and have allowed the 2 stores to grow in excess of 90GB in size (although with plenty of white space) and allowed the free space to drop to around 15GB. We had a failed drive on the box at the weekend and the RAID controller didn’t do too good a job of keeping things going resulting in corrupted stores/logs. We mounted empty stores to allow the affected users to continue working but now have to copy almost 200GB of data across the network to attempt a repair or restore (I believe strongly in keeping an original intact copy of the corrupt DBs/logs when trying any repair procedure). We are running a restore using a recovery storage group on an alternative server but until we have a safe copy of the old DBs on another box we cannot risk deleting the originals to make room for merging the data back.

We’re out of empty drive bays and the server doesn’t support USB so for now we’re just waiting for the copy to complete. 24 hours and counting…

[tags] Exchange, SAN, storage[/tags]

Posted by: Anthony Drewery | April 28, 2006

Reply/forward size limit when using public folders via OWA

The company I work for has been bought by another and we are currently transitioning our IT systems. As part of this transition the users are being given mailboxes in the new organisation. Until the decommisioning of existing systems is complete some users have a need to access their original mailbox and public folders which they are achieving via Outlook Web Access.

The users working with public folders via OWA found that they could not reply or forward messages exceeding 2MB in size. A Google search returned this article which provided the solution.

On each front-end server the following registry entry needs to be created: 

Location: HKLM\System\CurrentControlSet\Services\MSExchangeWeb\OWA
Value: maxPFReplyForwardSize
Type: REG_DWORD
Value Data: X

Where X is the maximum size you want to set in kilobytes (remembering to set the base to decimal).

[tags]Exchange 2003, OWA, Outlook Web Access[/tags] 

Posted by: Anthony Drewery | April 18, 2006

Failure of IPSec Services

I was working on a Small Business Server 2003 system at the weekend. My first job of the day was to install the latest security patches then reboot. Unfortunately after the reboot the server had dropped off the network and the IPSec Services service had failed to start. Manually trying to start it resulted in "The system cannot find the file specified". I tried another reboot for good measure but got the same result. I then disabled the service and rebooted. The server stayed on the network but components like Sharepoint and OWA would not work.

My initial suspicion was that this was caused by one of the security patches I had applied so I uninstalled them one at a time trying to find which was the culprit. However, after they all had been removed the issue still persisted. A bit of Googling the event logs showed that the IPSec Services were entering "block mode" and led me to the solution to my problem with this article:
http://support.microsoft.com/?kbid=912023. It seems that there was a corrupted file in the policy store. After resolving the issue I reapplied the security patches without any trouble.

You Had Me At EHLO has the solution to an issue we experienced when installing some of our Exchange 12 boxes. We got round it at the time by using the RC1 Refresh version of MMC 3.0 but we’ll use the registry key and RTM version in future.

You Had Me At EHLO… : Resolving MMC 3.0 errors when installing Exchange 12 CTP release

Posted by: Anthony Drewery | March 20, 2006

Exchange 12 Topology

I’m still trying to get my head around Exchange 12 mail flow and message routing but I’m sure it’ll become clearer once we get the servers up in the other sites. I’m hampered a little by the fact that this beta doesn’t have a GUI for configuring the bridgehead servers so I can’t have a nose around the settings.

From my understanding so far Exchange 12 is an Active Directory site aware application so it is very important that your Subnets and IP Site Links are correctly defined in AD Sites and Services. After checking available hardware in the various regions I’ve settled on 4 AD sites in the sandpit environment. These are Crawley (UK), Tampa (US), Santos (Brazil) and Montreal (Canada). The sites will be fully meshed with IP Site Links. To begin with the links will have uniform costs but at a later stage I’ll be investigating whether adjusting the values has any impact on mail routing within our organisation.

AD Site Overview
For the Crawley site we’ll have 4 servers initially. There will be a domain controller containing the global catalog, a mailbox server, a server with the bridgehead and client access roles and a gateway server. According to the release notes the gateway server role hasn’t completed the security pass. Therefore we’ll keep it on the internal network and route email to/from it via a Mailsweeper server in our DMZ.

I’ve placed a diagram of the Crawley topology below. Click on it for a larger version.

Crawley Topology

I’ll provide more information once the other sites are running and we’ve had a chance to experiment with connectors and mail routing. There’s no point me regurgitating the help files at this stage when I don’t have a decent understanding.
[tags]Exchange 12[/tags]

Posted by: Anthony Drewery | March 20, 2006

Exchange 12 – server naming conventions

Having had some time to review the Exchange 12 help files and documentation I’ve revised my infrastructure design slightly. In this post I’ll address server naming conventions and follow it with a post on my revised topology.

We use strict server naming conventions in our production environment and I’ve carried these through to our AD/Exchange 12 sandpit with some slight modifications.

I’m prefixing all server names with LAB. If for some reason they are picked up by one of our systems (e.g. LanDesk) it groups them together and makes it immediately obvious that they’re not production machines.

Next comes a three letter location code: MTL – Montreal, CRW – Crawley, TPA – Tampa etc.

That’s followed by EX. This makes it clear that they are Exchange servers and makes them easily identifiable to those who don’t work in the Messaging team and understand the role codes that follow.

Now we add a code that identifies the server’s primary role:

MB – mailbox server
BH – bridgehead server
CA – client access server
GW – gateway server
UM – unified messaging server

We might also use PF if the box was a dedicated public folder server.

Lastly we add a numeric increment to prevent duplicate names. We wouldn’t bother with this where you can only have one server of a particular role in a site e.g. SRS in Exchange 2003.
So the final server name would look like LABCRWEXMB2 or LABSYDEXUM1. We’re also using a similar convention for non-Exchange servers in our environment e.g. LABCRWDC1.

[tags]Exchange 12[/tags]

Posted by: Anthony Drewery | March 20, 2006

Exchange 12 installation screenshots

Here are some screenshots from the Exchange 12 installation process. Click on the images for a larger view.

E12 Installation 1 E12 Installation 2

E12 Installation 3 E12 Installation 4

E12 Installation 5 E12 Installation 6

[tags]Exchange 12[/tags]

Posted by: Anthony Drewery | March 17, 2006

Exchange 12 First Installation

Ok, our first Exchange 12 server is up and running. As this is a test environment and we are using slightly dated hardware we installed the 32 bit version. There is an excellent post at You Had Me At EHLO that talks about choosing Exchange 2003 hardware that can be reutilised for Exchange 12.

We installed on to a Windows 2003 R2 member server. This was part of a domain where we had raised the domain and forest functional levels to Windows Server 2003 Native. This is achieved via Active Directory Domains and Trusts. One of the pre-requisites in the Exchange 12 Release Notes is to have a domain functional level of Windows 2000 Native or greater. The default is Windows 2000 Mixed.

Other pre-requisites included Microsoft .Net Framework 2.0 (found under Optional in Windows Update) and Microsoft Management Console 3.0 (found here). We also assumed the Exchange 2003 pre-requisites of ASP.NET, IIS, SMTP and NNTP but our first attempt at installation failed because NNTP must not be installed.

After running setup and accepting the licence agreement we were presented with a choice of server roles. These were:

Bridgehead
Allows the routing between Active Directory Sites (it appears routing groups and admin groups have been dropped).

Mailbox Server
Provides the backend functionality including mailbox and public folder stores.

Gateway Server
Runs in your DMZ to provide ‘messaging hygiene’. I figure that’s virus scanning and spam filtering.

Client Access Server
Provides client connectivity for methods other than Microsoft Outlook e.g. OWA, ActiveSync.

Unified Messaging Server
Integration with the corporate phone system.

For this server we chose Mailbox Server as we have a second server to be built which will take the Bridgehead and Client Access roles.

I noticed that there were no /domainprep or /forestprep commands to be run. This appears to be covered in an installation step called Organisation Preparation.

So after 3 unattended steps (Copy Files, Organisation preparation and Server Roles) the installation was complete…

Note: I’ll update this post with some screen shots from our second installation.

[tags]Exchange 12[/tags]

Posted by: Anthony Drewery | March 17, 2006

Exchange 12

We’re just beginning a pilot/test implementation of Exchange 12 beta running on a global Windows 2003 R2 Active Directory. This is a sandpit environment separate from our production AD.

The initial infrastructure will have routing groups in the UK, US, Canada, Australia, Brazil and possibly New Zealand. Each site will start with a domain controller, front-end server and back-end server but this will expand as we manage to source additional hardware or make use of virtual servers. I hope to be able to add clusters and possibly storage area networks to the environment.

I’m installing the first Exchange server today so expect posts on Exchange 12, Active Directory and Monad over the next few weeks.

[tags]Exchange 12, Monad[/tags]

« Newer Posts - Older Posts »

Categories

Follow

Get every new post delivered to your Inbox.