This project has moved. For the latest updates, please go here.
1

Closed

Bridge Throws Error Instead Of Recovering When Web Service Throws Error On Deleted Message.

description

The web service throws an error when a NNTP client requests information on an article that has been deleted. The Community Bridge does not recognize this error and re-throws it, causing the NNTP client (e.g., Outlook Express or Forté Agent) to time out, and retry the operation - which only repeats the same sequence of steps that led to the error. This makes it impossible for a client to retrieve messages from the group that has had messages deleted. Attached file contains an example of the error being thrown.

file attachments

Closed Nov 13, 2011 at 9:11 PM by
There is now a hotfix with the work around. For more info see the latest patches:
http://communitybridge.codeplex.com/SourceControl/list/patches

Also a new release will be available soon which contains the workaround.

See also:
// 2011-11-13 V45
// - Workaround for Web-Service Exception (This item has been deleted)
// This might lead to slow responses of the bridge if you try to transfer the articles to the client and one article in a
// list of articles has this problem. Therefor this is only a work-around!!!
// MS need to solve the problem.
// Bug even the problem is solved, you can still use this new version, because the workaround will not hit anymore,
// if the web-service has resolved the bug
// For more info see also: http://communitybridge.codeplex.com/workitem/9553
// and http://social.Msdn.microsoft.com/Forums/en-US/reportabug/thread/dbe344b1-c0c2-4015-9e79-b975e06e0fe1#e060c569-a3c4-4744-ab24-10093f5cf241

comments

jkalmbach wrote Nov 13, 2011 at 1:13 PM

Can you provide an example forums and example message id!?
Please activate the logging the the bridge and put also the logifle here, so we can see the group and messageIds, which causes this problem!

JZeitlin wrote Nov 13, 2011 at 3:52 PM

The group in question is the Hey Scripting Guy forum, NNTP name Technet.en-US.ITCG. The error is thrown when the message numbers are being listed; my client (Forté Agent) attempts to get a full list of message numbers when an attempt to retrieve a message by a specific number fails, as can be seen in the attachment. Clearly, when a single message is requested, and does not exist, the Web Service handles this properly, returning error 423 - but not when listing the message numbers en masse. The attachment was generated by asking Forté Agent to retrieve all messages since 11/10/2011 00:00. Note that Agent attempts to do a binary search to locate the beginning of the group of messages to be retrieved, and then tries to get a complete list of messages when one of the messages that should exist fails to do so.

This suggests a possible technique to work around the error; when the missing message error is thrown by the Web Service, the bridge might attempt to retrieve the messages/headers from that group of 150 individually, and then report back to the client based on the results of the workaround, as though the Web Service error never happened. I do not know how practical it would be to implement the technique, however.

JZeitlin wrote Nov 13, 2011 at 4:38 PM

Attachment 3 (MSForumsErrorThrow3.txt) shows that the problem happens regardless of whether the messages are retrieved using LISTGROUP or XHDR; I had thought of the possibility that Agent would succeed in pulling the data via XHDR where LISTGROUP failed. I would guess at this point that the problem is connected with mishandling of database indexes in the Forum software and/or the Web Service.

(N.B. I have NOT trimmed this third attachment, in case the full response to the XHDR is required. The attachment referred to in my previous is Attachment 2, MSForumsErrorThrow2.txt. I will maintain the name, merely incrementing the number, as further attachments are requested/required.)

jkalmbach wrote Nov 13, 2011 at 6:24 PM

The best solution would be, that the web-service would only return the existing messages...

I can try to do this work-around, but it is only a work-around which would made the bridge slower in some situations...

jkalmbach wrote Nov 13, 2011 at 6:44 PM

I have now a short repor-step with the MS bridge:

telnet localhost 119

200 Server is ready and posting is allowed.
GROUP Technet.en-US.ITCG
211 51622 1 51622 Technet.en-US.ITCG
XOVER 7520-7521
224 Overview information follows
7520 =?utf-8?B?UmU6IExhc3QgbG9nIG9uIGZvciBhIGxpc3Qgb2Ygc3BlY2lmaWMgdXNlcnMgaW
4gQWN0aXZlIERpcmVjdG9yeQ==?= =?utf-8?B?RGVlcFNwcmluZw==?= Tue, 10 Nov 2009
17:25:07 +0000 0 0
7521 =?utf-8?B?UmU6IERldGVybWluZSBwYXNzd29yZCBjaGFuZ2UgaW5mbw==?= =?utf-8?
B?R3VubmVyOTk5?= Tue, 10 Nov 2009 17:42:26 +0000 0 0
.
XOVER 7520-7522
224 Overview information follows
503 Web service error



Hopefully MS will solve the problem with the web-service soon!

I wil try to implement an work-around...