CDR Tickets

Issue Number 3452
Summary Fix Program to FTP Vendor Data to FTP Server
Created 2011-11-14 15:11:50
Issue Type Improvement
Submitted By Englisch, Volker (NIH/NCI) [C]
Assigned To Englisch, Volker (NIH/NCI) [C]
Status Closed
Resolved 2012-03-09 12:20:11
Resolution Fixed
Path /home/bkline/backups/jira/ocecdr/issue.107780
Description

BZISSUE::5146
BZDATETIME::2011-11-14 15:11:50
BZCREATOR::Volker Englisch
BZASSIGNEE::Volker Englisch
BZQACONTACT::William Osei-Poku

Last weekend the program that runs as part of the publishing job which creates a compressed tar file and copies it to the CIPSFTP server failed to copy the data.
As a result, last weeks PDQ XML data has been extracted and copied to the public location.

We want to create a mechanism that submits a notification if this should happen again instead of silently failing.

Comment entered 2011-11-29 11:27:18 by Englisch, Volker (NIH/NCI) [C]

BZDATETIME::2011-11-29 11:27:18
BZCOMMENTOR::Volker Englisch
BZCOMMENT::1

This same problem came up again this weekend. After we haven't had any problems with this for at least the beginning of the year this problem came up three times within the past eight weeks.

We still want to create a notification if the error occurs again but the real issue is something else. Tar is reporting the error message:
tar: file changed as we read it

Bob/Alan: I know there exists an option to ignore this error in GNU-tar.
Are you aware of such an option with CYGWIN tar?

Comment entered 2011-11-29 12:37:20 by alan

BZDATETIME::2011-11-29 12:37:20
BZCOMMENTOR::Alan Meyer
BZCOMMENT::2

(In reply to comment #1)
...
> Bob/Alan: I know there exists an option to ignore this error in GNU-tar.
> Are you aware of such an option with CYGWIN tar?

The cygwin version is a port of the GNU tar, so I assume the option is the same. I think it would be:

--warning=no-file-changed

(see http://www.gnu.org/s/tar/manual/html_section/warnings.html)

I thought the idea of a "warning" was to issue the warning message but not stop processing, but I guess not if that's what's stopping us.

Do we know why the file is changing during the tar process?

Comment entered 2011-11-29 13:11:31 by Englisch, Volker (NIH/NCI) [C]

BZDATETIME::2011-11-29 13:11:31
BZCOMMENTOR::Volker Englisch
BZCOMMENT::3

I'm actually getting an error and an exit code 1.

> Do we know why the file is changing during the tar process?

The only thing that I can imagine would be that a backup is running changing the access time of a file.

Comment entered 2012-02-24 15:17:03 by Englisch, Volker (NIH/NCI) [C]

BZDATETIME::2012-02-24 15:17:03
BZCOMMENTOR::Volker Englisch
BZCOMMENT::4

So far, if the program failed I was writing a message to the log file. Now I have modified the program
FtpExportData.py - R10333
to also submit an email about a failure of the program so that it hopefully won't happen anymore that licensees are receiving the vendor data from the last week because the current data failed to be FTP'ed.

I've tested this change successfully on MAHLER by modifying a document that was in the process of being tar'ed. Since this was a fairly minor change (adding a few lines of text to be submitted by email) and it may take a long time before we "hit" this situation again to trigger the new code, I went ahead and copied the changes to BACH already.

I'll keep the issue open for a week or two before closing it.

Comment entered 2012-02-27 10:52:13 by Englisch, Volker (NIH/NCI) [C]

BZDATETIME::2012-02-27 10:52:13
BZCOMMENTOR::Volker Englisch
BZCOMMENT::5

(In reply to comment #4)
> I'll keep the issue open for a week or two before closing it.

There were no problems during the weekly publishing job.

Comment entered 2012-03-04 11:40:20 by Englisch, Volker (NIH/NCI) [C]

BZDATETIME::2012-03-04 11:40:20
BZCOMMENTOR::Volker Englisch
BZCOMMENT::6

Good news and bad news:
The good news is that the notification is working as implemented if the job fails.

The bad news is that the FTP job failed.

I'm closing this issue since the change is working properly but I may make additional changes if this job is going to fail more frequently due to this 'file changed as we read it' failure.
We may even decide to completely ignore the message. The last few times that we got this error and I recreated the tar file it turned out to be identical to the file that had been created but finished with a failure.

Comment entered 2012-03-09 12:20:11 by Englisch, Volker (NIH/NCI) [C]

BZDATETIME::2012-03-09 12:20:11
BZCOMMENTOR::Volker Englisch
BZCOMMENT::7

Closing issue. The program ran correctly for the past two weeks.

Elapsed: 0:00:00.001215