Hello, In my opinion, it will be easier because the database will be less loaded, but more importantly. I added wait_for_update in one method, which means that another scheduler will not be able to damage the state of the database. Most of these changes even streamline HA, because more things are done atomically/in one transaction.
I wonder what are the plans for HA. If this is based on distributing files to several nodes automatically, then this should not be a problem. These changes focus mainly on DagFileProcessor. This is a fairly new class (I proposed creating it), which contains code responsible for processing one file. Best regards, Kamil On Thu, Feb 27, 2020 at 6:36 AM Maxime Beauchemin < maximebeauche...@gmail.com> wrote: > Hey, I wanted to echo the awesomeness once more, but also bring up the > question as to whether any of this work may make it harder to distribute / > HA the scheduler down the line (?) > > I almost started analyzing the code and thought it'd just be easier to ask > the authors. > > Max > > On Wed, Feb 26, 2020 at 11:44 AM Felix Uellendall <felue...@pm.me.invalid> > wrote: > > > Really awesome indeed! I hadn’t enough time to look into all of it yet > but > > will definitely do. > > > > Thanks Polidea and Databand for your efforts you put into this! Kamil > > especially! > > > > -Felix > > > > Sent from ProtonMail Mobile > > > > On Wed, Feb 26, 2020 at 08:54, Sumit Maheshwari <sumeet.ma...@gmail.com> > > wrote: > > > > > Awesome work guys!! Kudos to all of you 👏 > > > > > > On Wed, Feb 26, 2020 at 6:59 AM Jiajie Zhong < > zhongjiajie...@hotmail.com > > > > > > wrote: > > > > > >> Good work! Thanks Kamil > > >> > > >> Best Wish > > >> — Jiajie >
