MSM Known Errors

Known Error 370, MAILSTORE Configuration Issue, Notification

Below are details of a recent Known Error identified by Marval UK which will, if implemented, offer a proactive solution to any of your customers who use multiple MOservers:
 
Symptom

Customers and assignees do not receive their notifications or they are dispatched a long time after being sent.

Setup Configuration

Multiple MOServer installation

V7.0.0.0 [build 1367 and higher]

Mail store configuration for each MOServer being a different folder (for example)

[NOTIFY]

MAILSTOREDIR  = c:\marval\server\mailstore

The Problem

When each MOServer is configured to use a different mail folder for the mailing thread, there is a tendency for the mail queue in the mail folder to remain in the folder for an undeterminable amount of time. 

The reason for this is regardless of how many MOServers are configured, only one mail spooler will be created.  The mail spooler is created when sending a notification and terminates when the mail store is empty or under an error condition.  If whilst the spooler is alive (initiated by MOServer one) sending emails a notification is sent through a second MOServer, the mail will not be picked up by the current spooler and this MOServer will not spawn an additional spooler.  So unless another notification is sent through the second MOServer mentioned here and there is no mail spooler, all notifications in the second MOServers mail queue will not be sent.

The Solution

To set the mail store folder for all MOServer in a cluster to the same folder, restart all MOServers together.  The process of moving from multiple to single mail store folders is not entirely without issue, because if there are notification files inside one or more of these folders, they will need to be moved into the new shared mail store.  Any file name conflicts will also have to be dealt with by renaming with different names, the convention consists of a numeric for the name and the extension of MVL

WARNING: When restarting the MOServer cluster, all other Marval services must also be restarted simultaneously.  Failure to do so may result in unexpected  behaviour.