Issue Number | 4095 |
---|---|
Summary | Prevent publishing software from setting bogus output path for push jobs |
Created | 2016-05-12 17:30:55 |
Issue Type | Bug |
Submitted By | Kline, Bob (NIH/NCI) [C] |
Assigned To | Englisch, Volker (NIH/NCI) [C] |
Status | Closed |
Resolved | 2016-12-19 16:12:27 |
Resolution | Fixed |
Path | /home/bkline/backups/jira/ocecdr/issue.184142 |
The publishing software populates the output_column of the pub_proc table with a non-empty path string, causing the software to write an error message when it tries to rename Job99999.InProgress to Job99999. Fix the publishing software so it doesn't do that.
The CDR server sets the value for the output_dir of the
pub_proc table for every publishing job. The value is first set
to an empty string (''), then the value is updated to the path of a
possible output directory. At the end of a push job to
gatekeeper this value for the directory is reset to an empty string
again.
The current code is trying to rename the directory name for export jobs
only assuming that only for those jobs the output_dir has been
set.
I have modified the publishing software to skip renaming the value for the output_dir if the job is a Push job. The question for ~bkline is if this solution is sufficient because it does not Prevent publishing software from setting bogus output path for push jobs.
The changes are currently on DEV.
R14406: cdrpub.py
A solution which uses a stored procedure would be able to prevent the publishing software from setting any output path for push jobs. However, the solution you have implemented is perfectly acceptable for the purposed of this ticket.
Note: This ticket can't be tested by the users. A developer needs to inspect the log file of a publishing/push job.
The error message
: renaming u'd:/cdr/Output/JobNNNNN.InProcess' to u'd:/cdr/Output/JobNNNNN':
Job NNNNN[Error 2] The system cannot find the file specified
has not been logged on QA since December 19th.
The software must have been copied to QA separately - prior to Bob's build, probably because publishing was turned off for a while on DEV due to WCMS deployments.
The software must have been copied to QA separately - prior to Bob's build ....
Right. That was necessary in order to test the work for OCECDR-4125 (fix bug in build/deploy software). It also helped (but wasn't essential) for testing OCECDR-4101 (tools for comparing server configuration between tiers).
Elapsed: 0:00:00.001463