- Article Type: General
- Product: Primo
- Product Version: 2
Occasionally a pipe run is cancelled due to maintenance work, etc. and there are multiple extracts that may contain the same records. A frequent question is does the Primo pipe know to harvest the .tar.gz files in order from oldest to newest (according to the timestamp).
Primo does harvest the files from oldest to newest, however there are multiple threads running in parallel to speed up harvesting and NEP processes. It is possible that since the files are normalized in parallel a newer one is completed before an older one, and as a result the DB is updated with the older info. This should happen in only rare cases. The only real way to prevent this from occuring would be to define that threads will not run in parallel – which will cause a pipe to take a very long time to run.
In case it is discovered that the old information is updated instead of the new information, it is advisable to run a new extract from the time that the pipe was not run, and to run the pipe again on the new extract.
- Article last edited: 10/8/2013