Managed Availability is able to monitor, detect, and fix problems that occur on an Exchange 2013 server. Its major benefit is reported as the ability for the Exchange developers to sleep well at night without being disturbed by reports of failures in the Office 365 datacenters. But it can be chatty, especially in the form of synthetic messages generated to measure database health that have to be captured alongside more important email if you journal on a per-database basis. Microsoft has a fix in Exchange 2013 - disable the probe!
Fellow MVP Jeff Guillet was the first to comment about the effect of the synthetic messages generated by the Exchange 2013 Managed Availability system on journaling. At that time we were still learning about the way that worked in production and some problems could be expected, especially around the finer operational aspects of the new server.
The problem doesn’t exist if you opt for per-mailbox journaling as obviously Exchange then only captures messages sent by the mailboxes you have selected for journaling. It does if you use per-database (or standard) journaling as this means that every message generated by the mailboxes in the database are captured, including those sent by the health (or monitoring) mailboxes that Exchange 2013 creates automatically in every mailbox database. Standard journaling is included in the basic Exchange license. Per-mailbox journaling is a premium feature and requires an enterprise CAL for every mailbox that is journaled. Jeff’s solution to the problem was to create a dynamic distribution group to specify the mailboxes to be journaled.
Now Microsoft has released KB2823959 to explain how you can manage the journaling of Managed Availability messages. Essentially, two options exist. First, you can elect to use per-mailbox journaling. This option requires more management oversight to ensure that the necessary traffic is captured and requires those enterprise CALs. For these reasons, you might be tempted to go with the other option, which is to create a global override to instruct Managed Availability to disable the probes that generate the messages and then restart the Microsoft Exchange Diagnostics service and Microsoft Exchange Health Manager service on all servers.
On the upside, establishing an override appears to be the right course to take as it fixes the problem and frees per-database journaling from the overhead created by the probe messages. Arguing against myself, disabling probes that presumably were implemented for a very good reason seems to weaken the ability of Managed Availability to monitor, detect, and solve problems that occur on an Exchange 2013 server.
Given that the software build number (620.29 for Exchange 2013 CU1) is included in the override command, there’s a fair chance that this exercise will have to be repeated when CU2 arrives, or perhaps Microsoft will have come up with a more elegant solution then. We can but hope.
On another topic, a candidate for “most useful blog post of the week” (I just made that award up but might well use it again) is the explanation by MVP Paul Cunningham of how an array of CAS (rather than a CAS array – a feature deprecated in Exchange 2013) work to provide better resilience for Outlook clients than relying on the failover capability from one CAS to another. Paul uses DNS Round Robin to illustrate how things work and makes the point that a load balancer is the way to do things properly in production. It’s a good example of clarification through simplification. Recommended!
Finally, Microsoft has released their Exchange 2013 Architecture poster, which is worthwhile downloading and printing, even on an A4 or letter-size page (readible with difficulty at some points). Copies were given out at TechEd North America and will be available at TechEd Europe.
Follow Tony @12Knocksinna