BlackBerry Forums Support Community               

Closed Thread
 
LinkBack Thread Tools
Old 02-24-2006, 08:40 AM   #1 (permalink)
bkw
New Member
 
Join Date: Dec 2005
Location: nyc
Model: 8800
Carrier: Cingular
Posts: 8
Post Thanks: 0
Thanked 0 Times in 0 Posts
Exclamation Blackberry service degradation when one Exchange server is taken offline..

Please Login to Remove!

Hey all, hoping someone may be able to assist or shed some light into a situation we're having.

We have Exchange servers clustered globally, and behind every bridgehead server we have at least one BES 3.6 server.

Whenever we perform a DR test and knock out one of the Exchange clusters, the performance of Blackberries worsens considerably.

For example-
We have exchange servers located in Chicago, NY and LA. We have BES boxes in each of these locations. We isolate Chicago, thus taking down Chicago's exchange and bes servers. What we're experiencing is that users (Handheld @ 4.0) on the LA and NY BES are seeing delays of up to 4-6 hours per 24 hours of site isolation! If we isolate a site for a weekend (48 hours), our users on the NY and LA BES have seen outages of 12 hours or more.

And yes, once everything is brought back up, mail routes normally and quickly.

Anyone else see anything like this? We've contacted RIM, and they seem dumbfounded.

Thanks!!
Offline  
Old 02-24-2006, 02:05 PM   #2 (permalink)
Knows Where the Search Button Is
 
Join Date: Nov 2005
Location: Washington, DC
Model: 7520
Posts: 25
Post Thanks: 0
Thanked 0 Times in 0 Posts
Default

Are you using a single BESAdmin account or is there a dedicated BESAdmin-type account for each physical site?
Offline  
Old 02-24-2006, 02:16 PM   #3 (permalink)
bkw
New Member
 
Join Date: Dec 2005
Location: nyc
Model: 8800
Carrier: Cingular
Posts: 8
Post Thanks: 0
Thanked 0 Times in 0 Posts
Default

Single BESAdmin account.
Offline  
Old 02-28-2006, 03:23 AM   #4 (permalink)
New Member
 
Join Date: Oct 2005
Location: London UK
Model: 8700
Posts: 13
Post Thanks: 0
Thanked 0 Times in 0 Posts
Default

Are there any BES users in NY or LA that point to the Chicago Exchange enviroment?. When users on a BES have their Exchange server offline the BES still tries to make connections to that Exchange server and normally throws back blocked threads thus causing email delivery problems. Have you checked the Logs?
Offline  
Old 03-03-2006, 11:01 PM   #5 (permalink)
Knows Where the Search Button Is
 
Join Date: Nov 2005
Model: 7290
Posts: 32
Post Thanks: 0
Thanked 0 Times in 0 Posts
Default

If you have users on a BES(especially 3.6) that have a mailbox on an Exchange that is down, info store offline, doing a backup, a defrag, you're having a high load and slow response....any problems between the BES and a related Exchange server, you are more than likely going to have mail delivery issues.

If you are going to be doing scheduled maintenance on an exchange, you could prevent mail delivery issues by moving the users to another Exchange server before (scanGAL should pickup the user moved, but you can run handheldcleanup -u for fun)
Offline  
Old 03-04-2006, 01:58 AM   #6 (permalink)
BlackBerry God
 
jibi's Avatar
 
Join Date: Oct 2004
Location: Jibi's Secret Place
Model: 8900
OS: 4.6.1.174
Carrier: AT&T
Posts: 11,310
Post Thanks: 0
Thanked 1 Time in 1 Post
Default

as bla said, this is a MAJOR issue with 3.6 that was pretty much resolved with 4.0 (unless, of course, the server/store down was the one the service account lived on)...
__________________
In the beginning the Universe was created. This has made a lot of people very angry and is widely regarded as a bad move.
Offline  
Old 03-04-2006, 12:58 PM   #7 (permalink)
New Member
 
Join Date: Dec 2005
Model: 7250
Posts: 10
Post Thanks: 0
Thanked 0 Times in 0 Posts
Default

He had the same issue until we moved up to BES 4.0 SP2.

BES 4.0 and BES 4.0 SP1 (and your version BES 3.6) certainly had the issue.

But we also moved a number of VIP users from remote Exchange servers to Exchange servers co-located on the same LAN as the BES servers.

Jason


Quote:
Originally Posted by bkw
Hey all, hoping someone may be able to assist or shed some light into a situation we're having.

We have Exchange servers clustered globally, and behind every bridgehead server we have at least one BES 3.6 server.

Whenever we perform a DR test and knock out one of the Exchange clusters, the performance of Blackberries worsens considerably.

For example-
We have exchange servers located in Chicago, NY and LA. We have BES boxes in each of these locations. We isolate Chicago, thus taking down Chicago's exchange and bes servers. What we're experiencing is that users (Handheld @ 4.0) on the LA and NY BES are seeing delays of up to 4-6 hours per 24 hours of site isolation! If we isolate a site for a weekend (48 hours), our users on the NY and LA BES have seen outages of 12 hours or more.

And yes, once everything is brought back up, mail routes normally and quickly.

Anyone else see anything like this? We've contacted RIM, and they seem dumbfounded.

Thanks!!
Offline  
Old 03-04-2006, 01:37 PM   #8 (permalink)
Knows Where the Search Button Is
 
Join Date: Nov 2005
Model: 7290
Posts: 32
Post Thanks: 0
Thanked 0 Times in 0 Posts
Default

4.0 may also still experience the issue(although it does deal with Exchange down much better), but it will not be as appartent (I'll explain). 3.6 would have a single, or multiple instances and (obviously) you will assign users to each instance as you do. If Exchange goes down, users on instances where a down Exchange user is located will be effected.
In 4.0, there can be up to 5 agents (by default anyways). These agents are somewhat similar to having multiple instances. Think of 5 agents, being like 5 instances in 3.6(althouth you can only have 4 in 3.6-blablabla), but one SRP and the Dispatcher will optimize the user/agent allocation according to (among other things) Exchange server location. So, if users from Exchange server1 are on agent2 and Server1 is down, those users will be effected. The controller will see a certain amount of wait counts (5 by default) and restart the agent, continuing up to 10 times (by default) in 24 hours (in which time you will need to restart).

How does the service account come into play?...
->if the Exchange where the service account is goes down, it should only effect users on the agent where users on that same Exchange reside.

Why?...the BES doesn't go back to that "admin" mailbox for any reason other than agent initialization. The reason why agent initialization needs to hit that mailbox is to grab a GC referal by making a mapi session (which happens for each agent in the same way)

What will the outcome be if the Exchange where the besadmin is located goes down?
->Any agent that gets restarted(for any reason, manually or by the controller) will not start. They will hit the max of 10 restarts very quick and you will need to restart the BES (check the CTRL logs for this.)

Wanna test it?...
have two Exchange servers, server1 has besadmin and nothing else, server2 has users.
start the BES and dismount the info store where bes admin is located. Email will still work (so will calendar and whatnot)

**this won't take all situations into consideration, but generally is true.
Offline  
Old 03-27-2006, 10:39 AM   #9 (permalink)
Thumbs Must Hurt
 
Join Date: Mar 2006
Model: 8700c
Carrier: Cingular
Posts: 63
Post Thanks: 0
Thanked 0 Times in 0 Posts
Default

We have a similar set-up. When I know that the remote site is going offline, I go into the BES admin console and disable email redirection for the users' at that remote site. My other BES users' are not affected.
Offline  
Closed Thread


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On





Copyright 2004-2014 BlackBerryForums.com.
The names RIM and BlackBerry are registered Trademarks of BlackBerry Inc.