Server Failure Notice

[pesphil]

Clients are using Outlook 2003 on XPPro workstations and after recently upgrading connector to v6.0.5737.3 they get many emails saying:

Server Failures were detected. No action is required.
This e-mail was generated for technical support purposes.
Please see the attached files for more details about the problems encountered.
Computer Name: PCname
Version: 6.0.5737.3
Parent Folder EID: 000000003efedffa70bc5d4cae91384c18f52362c2800000
Parent Folder Name: Inbox
Parent Folder ID: 2
Target Object EID: 000000003efedffa70bc5d4cae91384c18f5236224bc2a00
Store: Zimbra - Username

I have attached a sample of the xml error report from one of these emails, they are all similar to this one, something about SaveDraftRequest?

This has been going on for several weeks now. I tried deleting and recreating the Outlook profile on one of the users and re-syncing with Zimbra server but still getting the same server failure emails. Any assistance would be appreciated.

...Phil

Outlook Connector version 6.0.3_GA_1912_6.0.5737.3
Zimbra server version 6.0.3_GA_1915.UBUNTU8_64 NETWORK edition

[fsiegel]

Questions:

1) Does this affect everyone, or just a few users?

2) Do these computers have any weird Outlook addins installed? We had a bug (40948, fixed in 6.0.2) where the Linked In addin was making strange calls that caused us to get this error. That one has been fixed, but perhaps there is another addin that is doing something we can't handle.

3) This is coming from the SaveDrafts request. Do you have a set of steps that reproduces this error? What are you doing that causes the drafts folder come into play? Autosave, explicit save, etc.?

[fselendic]

Happens to us also. When it starts, it never stops. We have three users with same problem, they get virtually thousands of those messages per day.

Of course, 2 out of 3 are CEOs of their companies, one can't fool Murphy's law.

The only remedy is to shut down "Errors to Inbox". ZCS 6.0.4 with patched ZCO. Contacted Zimbra support, but it takes weeks, and most that I've got for now is "that's something with Draft folder", and "can you reproduce steps, can you enable logging".

I can't reproduce steps, it happens all the time. And am having trouble getting the logs, those are remote uses (different countries), and they are rather, huh, well, hostile. I would be also if 150 messages per minute are flooding my Inbox. Especially if those weren't the only fantom messages they get.

We also have one nice from System Admin: Undeliverable: Not Read. <Message title>.

That one happens A LOT when importing their old messages from PST (months and even years old), and also during regular day-to-day work, when everything is imported, and messages from some particular people come in.

Guys, what did you do with ZCO?
Was rock solid back in 5.0.16 days.

[fsiegel]

Well, since we never get this error here, it's kind of hard to help you if you can't give me reproducible steps and you can't give me logs.

You say you have 3 users with the same problem. How many total users do you have?

If it's only 3 out of a bunch, then I think there is some Outlook addin or some configuration of these machines that is causing problems for the connector. I'm not sure what I can do except to ask you to see if you could tell what is different about these 3 computers. Perhaps a sysinfo.

And are you sure that if these users ran ZCO 5.0.16, they wouldn't get the same issue? You should be able to run ZCO 5.0.16 against a 6.0 server. Maybe that's worth a try, even though I realize that CEOs aren't especially big on uninstalling and reinstalling connectors. I can get you that connector if you need it.

[fselendic]

Well, since we never get this error here, it's kind of hard to help you if you can't give me reproducible steps and you can't give me logs.

You say you have 3 users with the same problem. How many total users do you have?

If it's only 3 out of a bunch, then I think there is some Outlook addin or some configuration of these machines that is causing problems for the connector. I'm not sure what I can do except to ask you to see if you could tell what is different about these 3 computers. Perhaps a sysinfo.

And are you sure that if these users ran ZCO 5.0.16, they wouldn't get the same issue? You should be able to run ZCO 5.0.16 against a 6.0 server. Maybe that's worth a try, even though I realize that CEOs aren't especially big on uninstalling and reinstalling connectors. I can get you that connector if you need it.

I understand that, I'm trying to get remote access to the non-CEO user, so that I can collect some logs.

I have that good 5.0.16 connector, not sure if it would help, though, will first try with logs.

What should I get? Can I post it here somewhere?

[fsiegel]

Enable ZCO logging (just logging and http logging), do as small a run as possible where you get the error, and post the log files here. Hopefully they won't be too big. Also, if you can get the snippet of the server's mailbox.log for the time of the error, that would help as well.

Also, I would like to know what (if any) Outlook these users are running. In Outlook 2003: Tools->Options->Other->Advanced Options->Add-In Manager. Then let me know what boxes are checked.

[fselendic]

Still no luck with getting logs.

But, now problems are getting bigger....seems that some emails are passing through, but without attachments. Can be seen in "Sent" on Outlook side, people can see email, can see that it is 2MB big, but no attachment?!

And some are Sent, but never delivered

What I've find out (I'll have a nice chat with sysadmin tomorrow) is that log is full of:

FileUploadServlet - File upload failed
org.apache.commons.fileupload.FileUploadBase$IOFil eUploadException: Processing of multipart/form-data request failed. null
....
Caused by: org.mortbay.jetty.EofException

and also lots of:

FileUploadServlet - File upload failed
org.apache.commons.fileupload.FileUploadBase$IOFil eUploadException: Processing of multipart/form-data request failed. timeout
....
Caused by: org.mortbay.jetty.EofException: timeout

Are those the same messages? Is that connecttimeout issue (60 sec by default, if memory serves me right)

[fsiegel]

Timeout in ZCO defaults to 60 seconds. You can make it larger by changing registry values under (HKLM\SOFTWARE\Zimbra\Connection.

Is it possible there is weird virus checking software on these offending machines?

[fselendic]

First I have to clear server logs of those upload failed messages. Hopefully it is just timeout.

[fsiegel]

Check this server bug: 40917. This says it was fixed in 6.0.2. I see you are using 6.0.3 server, so I'm not sure what's going on, but perhaps you should upgrade to 6.0.4 server and see if that fixes your issue.