andrijapanicsb commented on issue #3383: template: copy md5 mismatch URL: https://github.com/apache/cloudstack/pull/3383#issuecomment-575197839 @DaanHoogland not sure that one fixes anything for existing users - existing checksums in DB will be the one which we are reading and the problem will not be solved, right? (it does fix indeed new templates) 1. I liked the idea from @shwstppr as the temp solution - to ignore the MD5 sum given from the DB, since it's the archive md5sim, not the template itself (which somehow doesn't work, can we make it work?) 2. I also like the idea that we do ad-hoc md5sum calculation on the template BEFORE we start copying it to another zone (I.e. don't read from DB, since it useless in the current implementation) - but do it ad-hoc (which does have it's issues of stressing CPVM CPU more frequently than needed) Either way, this needs more serious change (4.15?) to implement what you have already done, but to also forbid users to enter md5sum of the archive - well, not forbid, perhaps that is used for the validation of the download... but then when extracted, we can overwrite that in DB with the md5sum of the extracted file (how about multi-disk-ova???). Another approach is to copy (to another zone) the original archive itself (assuming we don't wipe it after the extraction), and do no more changes. Here, keep in mind for VMware, I saw 2 x OVA files (the original one and the generated one), both archives by definition, and the new OVA (generated) is being copied over and unpacked again (well this doesnt work, bug...). For KVM not sure what is being copied over. I can rant until tomorrow - but I believe simply always keeping the archive/original file and have that one's md5sum in DB AND COPYING that one always (and unpacking as needed) is the way to go. Forget ever copying the extracted file (why would we do that - more space, slower, etc?)
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
