Good Morning Frank: If you are running rsync with the --partial option, in case of failure, you can rerun the same rsync command again when it fails and it will continue where it left off.
In practice the difference between the code and the data will be minimal. We do not change existing data schemas or their associated code interpretation. The only case where the code and data can diverge is when new types of data are introduced, or new features are added to display existing data schemas. I would guess the actual incidence of problems would be rare. If you keep your data rsync running until it is successful, you will catch up to any new code features. Recent weeks have been intense for the hgdownload server. A massive amount of data was released in September. Many weeks since then the server has been %100 busy 24x7 as everyone was trying to catch up. That business is calming down at this time. We are exploring options to increase the hardware capacity of the download server. Thank you for your loyal support of the genome browser. --Hiram Frank Sørensen wrote: > Hi All, > > We're having problems downoading data from hgdownload.cse.ucsc.edu. > > We run a weekly update every tuesday night at 00:02 UTC. Every once in a > while everything works fine and we have a fresh complete update. This > process takes a couple of days max. Unfortunately we experience most of > the times, that rsync crashes long before we get to update and rebuild > the kent source tree. > > The source tree update (via git) usually works just fine, but I wonder > how intimately are the code and data are entangled. Is it is safe to > rebuild the source tree even if the database update has crashed? As a > safety precaution, we always skip updating and building the source tree > if the database update fails, which often results in a lag in our genome > mirror version updates. > > Kind regards > > - Frank Sorensen _______________________________________________ Genome maillist - [email protected] https://lists.soe.ucsc.edu/mailman/listinfo/genome
