frenchdMemberOct 10, 2011 at 2:35 am #156463
Good morning guys
This is the first time I’ve posted on petri.co.il, I’ve read many forum posts in the past and appreciate the help you guys give.
I’ve currently got an open ticket with Microsoft regarding the problem in this post but after 4/5 weeks we’ve yet to resolve it – I thought I’d post here to see if anyone has seen this before or has any advice.
Basically we’ve run Windows XP SP3 and Outlook 2003 SP3 connected to Exchange 2003 SP2 for many years and it works fine. Outlook is in cached mode and our Exchange environment is externally hosted by IBM at their datacentre in the UK.
We started piloting Windows 7 x64 (SP1 + all updates) with Office 2003 (SP3) and under this configuration we’ve consistantly seen bad performance on the client. By bad performance I mean this: Outlook 2003 is opened and takes much longer than XP to connect and update all the folders. If you have any additional mailboxes added to your profile, then these can take upto 6 or 7 minutes to open and display the contents of the folders. The worst performance though is seen when you attempt to open another server mailbox folder (File > open > other users folder). This takes roughly 10 minutes and on XP it was done in seconds. Booking meetings where a resource has been added (normally a meeting room) also causes outlook to hang for a few minutes. Basically any client > server communication is causing the client to hang. Working on your local cache is fine and works normally.
So far Microsoft have done the basic troubleshooting to eliminate the client: Outlook in safe mode, Windows 7 in safe mode with networking, new profiles created, add-ins disabled. We know it is not a client issue because we have 40 or so machines in our Win7 pilot and they ALL suffer the exact same issues, are all fresh Win7 builds and are on different hardware paltforms (all Lenovo).
Through some network monitoring and hang dump tracing, Micrsoft confirmed the client hanging is a result of a RPC call from client to server not getting a response. They tell me that the RPC protocol was re-written in Win7 to make it more efficient and that is the difference between XP and Win7 and why we only see this behaviour under our Win7 machines not not under XP. The question is though, why do WE see it any nobody else (Microsoft have not seen this before apparently).
So where have we gone from here?
Microsoft pointed us at a KB article regarding black hole routers (KB314825) asked us to confirm if the MTU settings on all devices on the logical route to the Exchange server were all set to 1500. Last week we managed to get this confirmation, that all MTU settings in the path are set at 1500. There is however a Nokia firewall inside the IBM network where Exchange is hosted. This is physically located on the last hop before you hit the network segment which the exchange server sits on. If I do a loaded ping (1472) with the DF flag set to the exchange server, it fails at the firewall; but IBM tell me this is because the firewall discards ICMP packets above 576 bytes. Ping packets of 1472 with DF not set work fine right up until the firewall. We’ve seen the config from this firewall and can confirm the MTU setting is set at 1500 on all ports. Therefore it seems that MTU is not the cause : however we do believe this firewall is somehow causing our Win7 clients to hang but are not sure why. For some reason these PRC calls from Win7 are not getting back to the client.
I have this ticket still open and will continue to troubleshoot with Microsoft but if anyone has any advice, ideas or heard this before please let me know.
I will post any updates and progress we make incase this helps someone else.
Thanks for reading, sorry its so long!
You must be logged in to reply to this topic.