HOYS Citizen Science

Welcome to the HOYS Newsletter

Dear Observers,

after a significant time of outage, the database server is now back online. A big thank you to two of my colleagues (Tim Kinnear, James Urquhart) for their tireless help to get us back on track! You can again upload data and process the images into the database, as well as extract the light curve of your favourite stars(s).
A note of caution, please: If possible do not all rush to upload all the data you have taken over the last few weeks and start processing immediately. If you can wait a day or two, then this might help to prevent any potential overload of the system due to too many people using it at the same time. Please also remember that between midnight UK time and 1am, we are running the daily backup. Please do not use the server during that time for anything (upload, processing, light curve searches). While processes that are started during that time will most likely fail, they might also cause data to be written wrongly into the database or the backup. Finally, should you encounter any unexplained errors, please do not keep trying the same thing multiple times, but rather send us an email so we can try to fix, whatever is wrong.

A brief explanation what happened, how we fixed it and what improvements we have made:

As far as we can tell, during the first week in April one of the three hard disks in our raid system failed, while data were written into the database. This caused in-correctable errors in a very small number of database entries. Only three images were affected. This went un-noticed for several weeks, as it could only be noticed when someone tried to query a light curve that contained one of the broken entries.
We have now copied over all the data (with some restrictions - see below) and I will re-process the three broken images in due course. Hence, no data have been lost.
While transferring the data, we removed all data points with photometric uncertainties above 0.3mag, as well as saturated sources, because none of these will ever get used in any of our data analysis. Furthermore, data entries have been restricted to a more sensible number of significant figures, i.e. positions are restricted to an accuracy of 0.1 arcseconds. This has shrunk the size of the database significantly and will allow slightly faster access during light curve searches.
We also prepared the database to be cross-matched against the next Gaia data release. Once this is public (June 13th) we will look into doing this cross-match in the background. When this is done, we will take the data base off-line for a short amount of time to test and enable the search via this cross-matched table, which will speed up the light curve search by a huge amount. Once this is finished, we will in principle be able to implement an alert system, which automatically picks up newly variable stars.

Best Regards,
Dirk Froebrich
HOYS Citizen Science
facebook twitter youtube 
Email Marketing Powered by MailPoet