Good morning all, I have an Exchange 2007 server that houses a hand full of mailboxes. The users that connect to these mailboxes are using Outlook 2007 and throughout the day continuly recieve: "Event id: 26 Connection to Microsoft Exchange has been lost. Outlook will restore the connection when possible" and when this happens: "Event ID: 19 Rpc call (EcDoRpcExt) on transport (ncacn_ip_tcp) to server (exchange02.domain.COM) failed with error code (6) after waiting (0) ms; eeInfo (none)" at about the same time. In Outlook the balloon notification will pop up letting the users know that Outlook has lost its connection with the Exchange server. It only happens with mailboxes on this Excahnge server, I can move the mailboxes back to another server and the issue goes away for that mailbox. My RPC Averaged Latency, RPC Operations/Sec and RPC Requests are generally under what Microsoft's high thresholds are. There are 3 shared mailboxes on this server that 100 or so users connect to where their mailboxes are housed on another server. Other than these 3 shared mailboxes there are only 3 user mailboxes and all three receive these disconnects and errors all day long. Any ideas?

Can u run check-mapiconnectivity test on that server. Next is the error apearing when the users are using RPC over HTTPS or when their are pure MAPI clinents. Are the MBX servers all in the same subnet, have u checked that RPC servive is working on the problematic server. Are the RPC ports open for the users to connect to the server. Are all your users on that server having the same problem, its again me asking if their connectiong like MAPI or HTTPS/RPC. Also run ExBPA and check for errors. Are their any system errors on the server indicating a hardware error?

Pure MAPI no RPC over HTTPS. Yes, all user maliboxes on that server have the issue. Yes, all mailbox servers are on the same subnet, connected to the same switch. RPC service is working on the server and as far as I know all default RPC ports are open. No system errors in the event viewer. ExBPA looks good. However, the Exchange Performance Troubleshooter indicated that there was a disk bottleneck found: "The ratio of Reads I/Os to Writes I/Os on drive F: was 341. Generally, this ratio should be less than 0.10; a higher ratio indicates that this disk is used for something other than writing to the transaction log files. For best performance, the log drive should be dedicated for transaction log files." When I press CTLR and right-click the outlook icon in the taskbar I see for that server 1422/2 Req/Fail. CPU on the server is around 5% average.

when u check get-logonstatistcs for a specific user on that mailbox how many connections to the mailbox do they have? this could indicate a network issue if u can see that the users have established connections to the server. Do u see errors on the server in the app logs, u should maby get an error whit the message "Mapi session exceeded the maximum of 32 objects of type “session” " if its a network problem , where u get disconnected constancy. For what do u use drive F: on that server, is it the same use for the other mailbox servers? Are the servers configured the same, meaning do u have the log files on the same place for every server (expample:all the log files are on disk (F:) on every server, or do u have on some servers located on C: and on some on d: ect.)? is it a SCR, CCR ? For the ExBPA issue u have, knowing what is he F disk used for and if its used the same for the rest of the serves could be important.

The users have about 5 connections each. No errors in the event viewer, app or system logs. The F:drive is for the databases, E: drive is for log files. Both exchange servers are setup the same way. There are however log files on this F: drive for a couple of the shared mailboxes...I need to change their path to the E: drive, I am sure that is why the Performance Troubleshooter threw an error for the disk bottleneck. No clustering is used. Thanks.

Just to make it clear the problematic users dont have more than 5 connections with the get-logonstatistcs cmdlet, when they get disconected permanently? Does the client go to and offline state at one point, after multiple attempts to connect to the server?

Also u can use the netowrk monitor to see what is exactly happening with the requests that the clients are sending to the server. I could be that the packages are being lost for some reason, if u say that u are using the same switch in the same subnet, a presume that it isnt a problem on the network. The users can connect to the mailbox with Outlook, but they just get disconnected constantly right? U should move the transaction logs for the shared mailbox, i know that it shouldnt be the issue, but again having the servers the same narrows the problem, aslo have u run test-mapiconnectivity?

Yes, outlook disconnects for a couple of minutes, the ballon pops up stating that the connection to the exchange server could not be made and then after a couple of minutes outlook reconnects. No, they have about 5 connections each.

yes, thats right. They just randomly get disconnected. ok, yes I do need to move those Tlogs. No I haven't, can you tell me how to run test-mapiconnectivity? thanks.

Ok, sorry. I figured it out. Test-mapiconnectivity results showed: MailboxServer: myserver Database: mydatabase Result: success Latency(MS): 29 Error: blank

also, when I run the get-logonstatitstics the users all had around 5 connections, but the shared mailboxes each had a ton. Could this be my problem, to many connections to each shared mailbox all at the same time? These mailboxes are also labeled as 'user mailboxes'. Would it be in my best interest to convert these to an actual 'shared mailbox' ?

http://social.technet.microsoft.com/Forums/en-US/winserverNIS/thread/9da2abd2-18e7-4ab5-babc-a98a1f7e4a15 Maby this post will help a bit, from your previous posts it think its a network issue, on the server or on the network it self. My suggestion is to check if the two Mailbox servers are configured the same, and eliminate that first, also try Network Monitor 3.4 from the client workstation to see what is happening to the packages. just for case did u run the test-mapiconnectivity localy, or from a different server?

I ran it locally.

Run it from a remote exchange server test-mapiconnectivity -server exchange02.domain.COM

Ok, I will try this. Also, lasst night I moved the transaction log files from the F drive to the E drive seperating the tramsaction logs from the databases. Hopefully I can gain a little increased perfromance from this.

I'm sorry.....Imeant when I said that I ran it locally I meant that I had ran it from my desktop. Do you want me to try it on the actual exchange server?

No need, just to make it clear, u have specified that the server that u specified in the cmdlet wan the problematic one?

yes, that is correct....I specified the exchange server with the issue in the comand.

when this disconnect happens, if I press CTRL and right-click the outlook icon in the taskbar, it shows all my connectivity to exchange. while the disconnect happens and I am viewing this connection status, the displayed view of my exchange server blinks several times really really fast as it disconnects, reconnects, disconnects and reconnects until it finally stops. Then under the Req/Fail column it reads something like 8070/27. Does this give a clue into what is going on?

I think the problem is we have 100 users all accessing these shared mailboxes all at once and when they move items from their inboxes over to them this causes a RPC load issue on the server.....do you think that that would be a fair assesment?

I also want to note that when the problem gets bad enough I have to restart the MS Information Store to remedy it.

Its an issue with the RPC packages not being sent correcty if i can say it like that, or them being lost on some point in the network, thast why u are get disconnected and reconnected several times, and in the end u get disconnected. U get disconnected cause the user has made 32 simultanious connections to the server and that is the max by default, i know that u said that when u check the logonstatistics cmdlet it doesnt show more than 5 to 6 connections for the user that is disconnected at that point but that is the issue. 100 users connecting to one server shouldnt be and issue even if it the all are on one mailbox, but again im not sure about that. If those shared mailboxes are only hosted on that server then maby thats the issue. U should check that also, Having to restart the Information Store will just restart the RPC connections that the users made to the server. My suggestion would be to call Microsoft Support to investigate it an to find the cause. But my guess is that if u turn those users to RPC over HTTPS they wont be losing their connections. But again that not a solution, just a workaround.

Ok, thank you for your help!!