Outlook and Zimbra Connector not sending attachments

[BenG]

Further info on this:
FConil and I are seeing two behaviours, which is confusing the issue.
The first, is as he described, and all the client gets is an error during the send/receive. So at least in this case, we are seeing an error message.

But we were also getting a more subtle error like JCPA's, where it silently fails, and the only clue is repeated bandwidth upload spikes.
This is far more serious, because this is costing us bandwidth allocation, and lots of it, as well as client dissatisfaction because things are "just slow".

No we have managed to fix both of these by adjusting the ZCO registry parameters.

In this case client machine is a Windows 7 running Outlook 2007 SP2, and Eset Smart Security Firewall, Server at 6.0.4 and ZCO is 6.0.4 patched.
With ZCO reg settings at 600 seconds.
Reg Key settings:
[HKEY_LOCAL_MACHINE\SOFTWARE\Zimbra\Connection]
"ConnectTimeout"=dword:000927c0
"SendTimeout"=dword:000927c0
"ReceiveTimeout"=dword:000927c0
"OptionReceiveTimeout"=dword:000927c0



Send a 11MB (8MB unencoded) email, and it works as expected, at about 300Kb/s upload, it takes about 7 or 8 minutes.
Although it does seem to do this twice(ie 2x 8mins), once to send and a second to transfer into the sent items?
And during this period (ie while doing the uploads) no incoming mail arrives, but it floods in as soon as the upload is complete.

Then I delete the file and see another upload spike for about 8 minutes, so I assume this is transferring to the trash?

After that, small emails come and go fine, and no further traffic spikes, even when I empty the trash.
Again, pretty well expected:


Then Change Registry setting back to default 60 secs as per below, close Outlook, rinse and repeat.

Reg Key settings:
[HKEY_LOCAL_MACHINE\SOFTWARE\Zimbra\Connection]
"ConnectTimeout"=dword:0000ea60
"SendTimeout"=dword:0000ea60
"ReceiveTimeout"=dword:0000ea60
"OptionReceiveTimeout"=dword:0000ea60

This time, in the network monitor, after sending the same email it runs for about 3O seconds and then it pauses for a few minutes and tries again, then runs for approx 40 secs and stops again.
During this period, small emails come and go as normal. But the email is never sent, and there is no error.

Change registry back to 600 seconds, close Outlook (It complains that there is still an email in the outbox).
Open Outlook again.

Spike in outbound network traffic for about 2x8 minutes, which I assume is the 11MB email finally going out, and then life is back to normal.
So, for what ever reason one of the 60 second settings is the problem.

And I would speculate that it is only an issue when you attempt to send an email that is greater than your outbound bandwidth/s*60. But I have spent way too much time on this already, so will leave mine at 600 seconds for the moment.

Rgds Ben

[fsiegel]

Note that we are bumping up the connection defaults in 6.0.5 (bug 43536) to 15 minutes.

[BenG]

That sounds like a good option, but is it just masking the root issue?
What I mean is, was the issue caused by a mismatching of (Outlook-ZCO-OS-Router-Zimbra) timeouts, or are we just effectively increasing the maximum size of message that can be sent. ie 15x60xMaxium Upload Bandwidth?

If so then my concern would be that even 15 minutes would not be enough.
We have clients that travel a lot, and some use 3G dongles that roam onto GPRS, or use various other low bandwidth connections.

Plus this bug has been around for so long, so can I suggest we also improve the error reporting within outlook?

For the synch to fail silently is just so frustrating.

Ideally the error should be visible within Outlook, so why not generate an error report as an email that details the email(s) at fault(To-From/Subject), their ID, the error(Timeout) and perhaps the number of times it has failed?
This would at least give us a fighting chance of diagnosing this remotely for a customer.

Rgds Ben