I am having some strange behavior with our Exchange 2003 Std SP2 server. We have a single Exchange server environment. We had a problem with this server completely locking up last week during a backup job that did not involve Exchange, but rather some file system data on the same server. Lots of I/O related to the backup at a time when the server was busy with some of its own I/O intensinve processes. There was no choice but to pull the plug and restart the server "hard", so to speak. We've since moved the backup to a different time window, with no more lock ups thus far. We are using Backup Exec 9.1 at the moment.
Anyway, since the server came back up, we have not been able to back up certain of our Contacts when doing a mailbox level backup. We will repeatedly see this in the backup log for many, but not all Contacts:
Access denied to file Jane Doe [jdoe]ÃƒÂ¯Ã‚Â€Ã‚Â¦Top of Information StoreÃƒÂ¯Ã‚Â€Ã‚Â¦ContactsÃ
The item \\FS01\Microsoft Exchange Mailboxes\Jane Doe [jdoe]ÃƒÂ¯Ã‚Â€Ã‚Â¦Top of Information StoreÃƒÂ¯Ã‚Â€Ã‚Â¦ContactsÃ
red Friendly in use - skipped.
The contacts come from many mailboxes, and appear to be more or less the same list of Contacts each night (four nights so far). Those same contacts can be opened and used by users. When this occurs, the Exchange server logs a single 9646 error from MSExchangeIS. Based on the following KB article, I have increased the number of messages a MAPI client can open, but the first jump from 250 to 500 was not enough:
Even were I to reach a number that was enough, the fact remains that pre lockup, this was not necessary. I wonder if there is store corruption that is not significant enough to impair the functioning of the server, but enough to impact the backup. I've reviewed this article:
But it talks about these steps as a means to recover an unmountable store, etc. May be extreme in my circumstances. May be related or may not be, but we have one user who is having difficulty sending mail. I am not convinced yet that it isn't an issue with the MAPI profile on her machine, but we are still troubleshooting. In any event she had a few days of trouble free usage after the lock up event, so the timing is wrong for this to be related.
Any suggestions as to a reasonable approach to this issue are greatly appreciated.