BlackBerry® Enterprise Server Version 5.0

[summest]

I just setup BES 5.0.1 SP1 MR2 and have had it running for about 1 week.  The install was very smooth.  I had to fix some minor issues, but overall, the system is pretty reliable.

I do see one annoying problem where you try to send a message and about 25% to %50 of the time I receive a red X and have to resend.  Resend works every time.  All users on BES 5 encounter the same issue.

If we saw a red X every time we sent a message and resend would not work, I could understand the problem.

Why would a majority of the messages go through fine and some do not and have to be resent?

Again, new BES 5 works great on Windows 2008 SP2.

Is there a performance issue?  I have 4G of RAM with only 2 users.  Even with one user, I see the issue.

Thanks for any guidance,

[knottyrope]

That is not normal, any errors in the MAGT logs?

[summest]

Thanks for the assistance Knottyrope,

I do see this in the MAGT log:

[40700] (03/19 06:41:36.965):{0x16B8} {user@carnival.com} Receiving packet from device, size=154, TransactionId=-256944738, Tag=1668, content type=CMIME, cmd=0x3
[30112] (03/19 06:41:36.965):{0x16B8} {user@carnival.com} Receiving message from device, RefId=1872903262, Tag=1668, TransactionId=-256944738
[20265] (03/19 06:41:37.815):{0x16B8} {user@carnival.com} MAPIMailbox::RIMRecipstoMAPIRecips - ModifyRecipients (0x80040200) failed
[30147] (03/19 06:41:37.815):{0x16B8} {user@carnival.com} MAPIMailbox:end(ppMAPIMessage) Cleanup - HrGetOneProp (0x8004010f) failed
[20477] (03/19 06:41:37.914):{0x16B8} {user@carnival.com} Send() failed: ERR_FAIL, RefId=1872903262, Tag=1668
[40277] (03/19 06:41:37.914):{0x16B8} {user@carnival.com} Sending message error to device for message 1872903262
[40583] (03/19 06:41:37.914):{0x16B8} {user@carnival.com} Sending packet to device, Size=47, Tag=3614, TransactionId=-878542998
[40279] (03/19 06:41:37.914):{0x16B8} {user@carnival.com} SubmitToRelaySendQ, Tag=3614

I also get this alert when the failure occurs:

<N/A> | BlackBerry Messaging Agent MIAMIBES Agent 1 (Application Event Log on cclprdinfbes.carnival.com) | 03/19/2010 06:41:38  (AFFF4FFD) -> {user@carnival.com} Send() failed: ERR_FAIL, RefId=1872903262, Tag=1668

<N/A> | BlackBerry Messaging Agent MIAMIBES Agent 1 (Application Event Log on cclprdinfbes.carnival.com) | 03/19/2010 06:41:37  (AFFF4F29) -> {user@carnival.com} MAPIMailbox::RIMRecipstoMAPIRecips - ModifyRecipients (0x80040200) failed

Notes:

A resend always go through fine.

Thanks,

[knottyrope]

What version excahnge?

How is the exchange server configured in relation to drives and where OS, tlogs, mail store is installed.

What type of drive are you using?

Raid level?

How many Exchange users?

Any other apps or programs installed?

[chmajo · Accepted solution]

We had a similar problem.  Our Exchange 2010 CAS servers are sitting behind a hardware load balancer (F5) and the load balancer was timing out connections after 5 minutes (as designed).  As a workaround we forced the Windows CAS servers to send TCP keepalives every 4 minutes (default is 2 hours), which resolved the issue and no more red X.

[summest]

Thanks for the reply.  My issue was resolved another way, but yours may be better.

We had to edit the registry and modify profile settings for the MAPI profile.  The profile was pointing to the CAS load balancer name.  The engineer had me change it to a global catalog server and I have not seen the problem since.

Although, your method may be a better fix.

Thank you very much.

[summest]

Hello chmajo,

     How did you set the TCP Keepalives to every 4 minutes?

Thanks,

-Steve.

[bpierfy]

---

summest wrote:

Thanks for the reply.  My issue was resolved another way, but yours may be better.

We had to edit the registry and modify profile settings for the MAPI profile.  The profile was pointing to the CAS load balancer name.  The engineer had me change it to a global catalog server and I have not seen the problem since.

Although, your method may be a better fix.

Thank you very much.

---

I know this thread is old but I'm trying to understand if what you did would help me as well.

We have a load balancer but the BES and the CAS servers are all on the same subnet, thus not allowing us to use the same DNS that we point outlook clients to.  We can get around this by putting DNS on the BES server (or using host file), but we'd like a different way.

We'd like to be able to force the BES to go any one of two CAS servers, without picking the DNS from AD or the CAS array.  Is this possible?

[PSCArmstrSM]

We've got all of our Exchange 2010 BES domains pointed to one CAS member by way of a hostfile entry for the CAS array.  ex: "xxx.xxx.xxx.xxx   MAIL.DOMAIN.COM"