Hi Dominique, Am Dienstag, 30. Mai 2017, 12:21:39 CEST schrieb Dominique Chabord: > 2017-05-30 12:00 GMT+02:00 Axel Braun <[email protected]>: > > I'm a fan of using packages procided from your distribution for running a > > production environment. Local extentions can ba applied to a separate RPM > > and installed / updated to production once sufficiently tested. > > in this case what does your update procedure look like ? > > your updated specific RPM depends on the Trytond official packages, then > - stop Trytond > - save databases > - update the code with RPM > - update every database > - restart in production (and cross fingers ?)
If you have an update in a minor release (4.4.0 -> 4.4.1) it is mostly sufficient to stop the server and start after installing the new RPM. Done that many times without problems > my questions are : > - what is the roll-back procedure in case some problems appear at any step ? Worst case...boot into the previous snapshot (if you use btrfs) > - is there a dry-run possibility without stopping production ? Depends on the system landscape. If you have a 3 tier environment (Devel, Quality and production) you may copy production to quality first and try the upgrade in quality. Depending on the amount of extensions you have build yourself it may not run too smooth.... Then you need to fix your extensions until it works. To make sure your have transactional correctness an offline upgrade is the only chance IMHO. Better save than sorry in that case. > - do you know about real production cases where this kind of update is > proven ? ( gnu-health ?) For the dot-upgrades we've done that in a company in Hamburg (unless they went out of business - not for Tryton reasons). Cheers Axel -- You received this message because you are subscribed to the Google Groups "tryton" group. To view this discussion on the web visit https://groups.google.com/d/msgid/tryton/10933862.H8rhF9ijrL%40t520.axxite.internal.
