Windows 2008 / Exchange 2010 connection issues with workstations
Sitrep:
Since two weeks now, we're experiencing connectivity issues between workstations (XP Pro, 7 Pro and 8 Pro) and our Exchange 2010 Standard server (Windows 2008 R2 Standard). We're using a single domain and all systems are on 192.168.4.x range.
The problem is that infrequently the pc's lose their connection in Outlook with the mailserver. Due to cache mode they can still work. Sometimes the connection automatically restores, sometimes they really need to reboot the pc or at least restart Outlook.
Outlook is configured as RPC over HTTP, with both chechmarks set for slow/fast connections and Basic Authentication. Previously we had them connecting directly through MAPI. Because at first we assumed the number of concurrent connections was too high on the Exchange Server, we changed to RCP over HTTP as recommended by Microsoft.
This seemed to help at first, but now clients with RPC are still having problems.
The problem started when we changed old XP machines for Windows 8 machines. The XP machines also had Outlook 2010 so that has not changed.
Both workstations and servers are completely up to date as of today, including rollups and what not.
- We've been using VMWare ESXi 5.1 for some time now, without issues. The Exchange server resides on a vHost.
- There is no pattern to be found in losing the connection (EventID 26 in the application log on the pc's)
- Some users have it 4 times a day, others next to that user have no problems at all
- Some users have it on Monday but not on Tuesday etc, have it on Wednesday or only once a week
- Our Exchange store is 215GB and split up in two stores; 150GB / 65GB. We moved users around from one store to the other but this did not help.
- We've rebooted ALL switches and checked them for errors (managed switches), but there are none
- The Exchange server is also one of the two DC's we have, the other DC is a Windows 2003 server who is also DHCP. The AD functional level is 2003.
Just now we've noticed that when the Outlook disconnects from Exchange, we ALSO can not PING the mailserver. We can PING all other servers and the gateway, but not the Exchange server.
As soon as the connection is back in Outlook, the PING returns to normal. At the same time, the workstation next to the one that loses the PING has NO issues and PING continues as normal.
It seems as if the workstations are actively being denied access to the mailserver/domain controller at random points. We just can not figure out why. There is no anti-virus/anti-malware or firewall active. Due to the random nature, it seems we hit thresholds of some sort now and then. Because this has never been an issue before and the total number of pc's has not increased, we're staring in the dark right now.
Finally; we also use a Remote Desktop Server. Many users are logged on to that and they have no issues at all. Outlook is always connected. The problem is only with the workstations and not related to specific users as far as we can tell.
Any insights? I probably have not mentioned all there is to say, but we've already tried so many things we're drawing a blank.
Since two weeks now, we're experiencing connectivity issues between workstations (XP Pro, 7 Pro and 8 Pro) and our Exchange 2010 Standard server (Windows 2008 R2 Standard). We're using a single domain and all systems are on 192.168.4.x range.
The problem is that infrequently the pc's lose their connection in Outlook with the mailserver. Due to cache mode they can still work. Sometimes the connection automatically restores, sometimes they really need to reboot the pc or at least restart Outlook.
Outlook is configured as RPC over HTTP, with both chechmarks set for slow/fast connections and Basic Authentication. Previously we had them connecting directly through MAPI. Because at first we assumed the number of concurrent connections was too high on the Exchange Server, we changed to RCP over HTTP as recommended by Microsoft.
This seemed to help at first, but now clients with RPC are still having problems.
The problem started when we changed old XP machines for Windows 8 machines. The XP machines also had Outlook 2010 so that has not changed.
Both workstations and servers are completely up to date as of today, including rollups and what not.
- We've been using VMWare ESXi 5.1 for some time now, without issues. The Exchange server resides on a vHost.
- There is no pattern to be found in losing the connection (EventID 26 in the application log on the pc's)
- Some users have it 4 times a day, others next to that user have no problems at all
- Some users have it on Monday but not on Tuesday etc, have it on Wednesday or only once a week
- Our Exchange store is 215GB and split up in two stores; 150GB / 65GB. We moved users around from one store to the other but this did not help.
- We've rebooted ALL switches and checked them for errors (managed switches), but there are none
- The Exchange server is also one of the two DC's we have, the other DC is a Windows 2003 server who is also DHCP. The AD functional level is 2003.
Just now we've noticed that when the Outlook disconnects from Exchange, we ALSO can not PING the mailserver. We can PING all other servers and the gateway, but not the Exchange server.
As soon as the connection is back in Outlook, the PING returns to normal. At the same time, the workstation next to the one that loses the PING has NO issues and PING continues as normal.
It seems as if the workstations are actively being denied access to the mailserver/domain controller at random points. We just can not figure out why. There is no anti-virus/anti-malware or firewall active. Due to the random nature, it seems we hit thresholds of some sort now and then. Because this has never been an issue before and the total number of pc's has not increased, we're staring in the dark right now.
Finally; we also use a Remote Desktop Server. Many users are logged on to that and they have no issues at all. Outlook is always connected. The problem is only with the workstations and not related to specific users as far as we can tell.
Any insights? I probably have not mentioned all there is to say, but we've already tried so many things we're drawing a blank.
@jrhelgeson: We just changed it to VMXNET3 (where Exchange resides). It was set to E1000. On our other 2 vHosts it is already VMXNET3. So far we still have Outlook disconnecting, but with cache mode on it's not an immediate issue.
Exchange server 2010 / Windows 2008 R2 was built up on VMWare. It did reside on another vHost before, back when we had only 2. Now that we have a third, we migrated to the third vHost. There are no other servers present on that vHost. Unfortunately we cannot migrate to another vHost due to 1) disk space, and 2) vMotion only works in offline mode because we do not have central storage. This would mean a lot of downtime for our customer.
Thanks for your answer so far.
Exchange server 2010 / Windows 2008 R2 was built up on VMWare. It did reside on another vHost before, back when we had only 2. Now that we have a third, we migrated to the third vHost. There are no other servers present on that vHost. Unfortunately we cannot migrate to another vHost due to 1) disk space, and 2) vMotion only works in offline mode because we do not have central storage. This would mean a lot of downtime for our customer.
Thanks for your answer so far.
I ask because I have had horrible luck with any adapter other than the E1000 that matches exactly what you have described.
IF you are using E1000 adapters, I would log into the machine through the console, go into device manager and delete the adapter, scan for new hardware devices and let it re-install. Then reconfigure the static IP address on the network card.
Those are the two steps that have resolved 100% of the flaky issues I've ever had with NIC connectivity on VM hosts.
IF you are using E1000 adapters, I would log into the machine through the console, go into device manager and delete the adapter, scan for new hardware devices and let it re-install. Then reconfigure the static IP address on the network card.
Those are the two steps that have resolved 100% of the flaky issues I've ever had with NIC connectivity on VM hosts.
So to get things clear; did you have issues WITH the E1000 adapter, or with any OTHER adapter than E1000? We've just switched from E1000 to VMXNET3 because we believed VMXNET3 is better for Server 2012, and hopefully for 2008R2 too.
In other words, do you advise to go back to E1000, or stay on VMXNET3?
In other words, do you advise to go back to E1000, or stay on VMXNET3?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
We might follow up on that advice then. And indeed, all this time we were using the E1000 adapter when we had these issues. The issues still present themselves using the VMXNET3, but less frequent. Problem is that now all settings are set-up correctly on all clients (using HTTP for fast and slow connections), so we actually suspect the switching itself now. It's working without failures (by analyzing the managed switches), but some are are old; very old.
To be continued, also in regard to the E1000.
To be continued, also in regard to the E1000.
Wasn't the total solution, but helped to get us closer. Changing switched and turning on cachemode helped the rest.
It sounds like 'TCP/IP stack' problems/corruption.
Was this a P2V migration, or was this built up on VMware?