Home Page › Forums › BizTalk 2004 – BizTalk 2010 › Loosing messages from FTP
- This topic has 6 replies, 1 voice, and was last updated 9 years, 10 months ago by
community-content.
-
AuthorPosts
-
-
November 12, 2007 at 6:56 PM #18401
Hi All,
I’ve got a problem with receiving files using an FTP adapter.
Configuration:
– receive port with no mapping inside and receive location (FTP adapter
+ XMLReceive pipeline with AllowUnrecognizedMessage=true)
– send port 1 (FILE + PassThruTransmit) to put incoming messages into a local folder
– send port 2 and/or orchestration to process messagesEverything use to work fine but from time to time (once in some days) I receive an error “Root element is missing”. For those messages, message body was completely empty with size = 0 bytes. The same with files stored in local folder (see send port 1). But FTP log file shows file was transferred with correct file size (5K for instance). There wasn’t any error message in the http://FTP.log
Here is an extract from http://FTP.log:
> LIST *.xml
< 150 Opening data connection for /bin/ls.
< -rwxrws— 1 root sb 5102 Nov 12 10:43 20071112104317.xml
< 226 Transfer complete.
> TYPE I
< 200 Type set to I.
> PORT xx,xx,xx,xx,x
< 200 PORT command successful.
> RETR 20071112104317.xml
< 150 Opening data connection for 20071112104317.xml (5102 bytes).
< 226 Transfer complete.
> DELE 20071112104317.xml
< 250 DELE command successful.
> QUIT
< 221 Goodbye.Any ideas?
P.S. FTP is hosted on *NIX system and files are produced by a Legacy System.
Thanks,
Nick
-
November 13, 2007 at 3:47 PM #18411
Hi Nick,
Can you describe what your FTP Adapter configuration is like? I.e Is it clustered, single host instance? Do you have a local log file? I ask as we ran into two problems with the FTP client that prompted hot fixes. If you symptoms are simlar to ours you may have impacted the same way.
For us we are using a Clustered Host and were storing the log file on a clustered resource drive. That was a problem, if it is on a local drive then it is ok. We also had problems where content was being stripped (zero byte files). We would try to receive 600 10k files and would regularly find that 1 or 2 of them would have its content stripped. For that issue we were issued a private hot fix.
We are also using Unix
Kent
-
November 13, 2007 at 5:11 PM #18412
Hi Kent,
Thanks for the reply.
Our configuration is quite simple in this case. It’s single host instance, no clusters, local ftp-log file. And yes, from time to time we’ve got stripped content (zero byte file). But did your log show correct file size when you received a zero-byte result? That’s what it shows for us and I wonder how can it be.
How did you fix the problem?
Regards,
Nick
-
November 13, 2007 at 6:19 PM #18414
Hi Nick,
I was just going through my notes…it has been a busy past 6 months. One of the symptoms that we had was that our receive location would just lock up. We would get a zero byte file in the FTP Temp folder and the receive location would still be enabled. It was much like a deadlock. We were issued a private hot fix for this but that does not sound like the same issue you are talking about. I don’t remember what the log file said and I can’t find it in my notes…sorry. So we contacted Microsoft Premiere Support and they issued us a hot fix. I don’t have a “fix number” just our case number.
Since you are not using a clustered host our other scenario does not apply to you.
Something else to consider is that when you are picking up these Files, the FTP adapter does not respect file locks from your legacy system. When performing load tests, we had a test harness that would load up the Unix directory with Files. We would leave the receive location on and would find that the occaisional partial file was being delivered. I do not recall what was in the FTP logs. Does this scenario apply to you? If so let me know and I can share how we got around this.
Kent
-
November 13, 2007 at 9:54 PM #18415
Kent,
Yes, lack of locking implementation in FTP/UNIX scenario is my major supposition at the moment. But http://ftp.log confuses me in that theory.
I communicate with support team of the legacy system trying to find out any issues on their side or review the files creating script. The idea is to use file with temporary name in the same folder while writing the content, and just rename the ready to go file from temporary name to the appropriate one.
I would try everything I could to resolve the issue. Your help would be highly appreciated.
Thanks,
Nick
-
November 14, 2007 at 6:27 AM #18419
Hi Nick,
This how we solved the problem, check out the 4th post in this link:
http://forums.microsoft.com/MSDN/ShowPost.aspx?PostID=2316907&SiteID=1
For us we have the “work” file and a “sig” file. The sig file is our trigger that lets the middleware(or in this case BizTalk) that a work file is ready for consumption. So we have a script that is scouring the sig folder every 5 minutes. When it finds a sig file it will look for the partner work file in the work folder. The script will then append a “.BIZ” onto the work file and will delete the sig file. We then have BizTalk looking for “.BIZ” files. With this scenario we have not had any problems with BizTalk consuming partial files.
Ken
-
November 14, 2007 at 2:32 PM #18426
Thanks Kent.
I will let you know when we resolve the issue.
Nick
-
-
-
-
-
-
-
AuthorPosts
- The forum ‘BizTalk 2004 – BizTalk 2010’ is closed to new topics and replies.