This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
2023-07-11
Channels
- # announcements (6)
- # babashka (35)
- # beginners (6)
- # biff (2)
- # calva (1)
- # cider (16)
- # clj-kondo (14)
- # clojure (176)
- # clojure-austin (13)
- # clojure-europe (7)
- # clojure-norway (6)
- # clojure-uk (12)
- # clojurescript (4)
- # cursive (11)
- # data-science (3)
- # datalog (10)
- # datomic (45)
- # events (1)
- # fulcro (2)
- # hyperfiddle (29)
- # leiningen (14)
- # lsp (2)
- # meander (2)
- # off-topic (24)
- # polylith (5)
- # re-frame (2)
- # shadow-cljs (21)
- # specter (14)
- # tools-deps (16)
- # xtdb (5)
Hello. We added a new XTDB node to our cluster, total of two nodes now. The two nodes only know each other through the common postgres backend. The new node is syncing/re-ingesting by the look of the CPU/mem. Each node has a separate DNS/Url. When connecting with xtdb/new-api-client to the old client and making queries it fails intermittedly with :xtdb.error/error-type :node-out-of-sync, :xtdb.error/message "Node out of sync exceptions. Is anyone able to explain what is going on?
I'm surprised that the addition of a second node has effects on the existing ones. I have to infer that re-indexing is not a read-only operation on the database. Or I'm doing something silly.
Something silly it is 🙃 apparently misconfigured alb and it was load balancing between new and old
hey @U054W022G, glad to hear you got to the bottom of it 🙂 are there any non-sensitive details you could share with us so that others don't do the same?
It depends a bit on the selected AWS topology. In our case, 1 ECS cluster, one ECS service for each XTDB instance, one ALB each service, one route 53 for each ALBs. Just be careful naming all correctly so the dns name routes to each target group. The goal here is to have writes into XTDB separated from querying to avoid fluctuation of reading performance when we need to transact a lot of data.