This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
2023-01-18
Channels
- # announcements (19)
- # babashka (30)
- # beginners (51)
- # biff (12)
- # calva (20)
- # cider (27)
- # clara (20)
- # clj-kondo (19)
- # clojure (128)
- # clojure-canada (1)
- # clojure-dev (5)
- # clojure-europe (29)
- # clojure-nl (1)
- # clojure-norway (48)
- # clojure-uk (5)
- # clojurescript (6)
- # clr (37)
- # core-async (17)
- # cursive (2)
- # data-oriented-programming (2)
- # datahike (1)
- # datomic (5)
- # emacs (22)
- # exercism (1)
- # funcool (17)
- # humbleui (15)
- # jobs (1)
- # malli (44)
- # off-topic (11)
- # podcasts-discuss (1)
- # pomegranate (1)
- # react (1)
- # remote-jobs (1)
- # shadow-cljs (25)
- # tools-build (12)
- # tools-deps (70)
- # xtdb (24)
We are using Datomic on-prem, and run the backups frequently so that we have many points in time to pick and restore from. We are considering running the backups as frequently as possible (in a loop essentially), and wondered what backup frequency others are using and if anyone knows of reasons we shouldn't.
We run the backup in a loop; but eventually the backups get too big (there’s no way to remove a backup point and GC the segments it uses that nothing else uses). So we do a full backup weekly, and while the full backup is running (because it takes a few hours) the incremental backup is also running (so we don’t have gaps).
Ahh, that's a really good point, and one I hadn't thought too much about as our backups are currently tiny. Thanks.
Heh, thinking about retention policy when dealing with an immutable database feels odd. I think that 2-4 weeks of high-frequency incremental backups should cover us. With longer term backups that we keep for as long as possible. I could look at running multiple backups at different frequencies of incremental backup? eg. 1 backup on loop that is reset every month, and then another that only runs every week that we archive and reset yearly, etc.