I'm guessing that's not a good fit. A pod doesn't persist state between failure. The guess here is that a transactor needs to hold state.

this would be relevent https://docs.datomic.com/on-prem/operation/ha.html

give you can run two transactors to provide HA then maybe a pod is fine, and you could get HA by using datomic pro.

I'm reeling at the idea of trying to make that work vs just using datomic cloud though.

I'm pretty sure the transactor state is persisted in the underlying storage

For development we run both the transactor and the peer in docker containers. It seems to work; but it's not a production environment at all.

I’ve seen transactors run in pods in production.

Cool, thanks y'all... I think the only concern is probably ensuring that it gets enough RAM. To me it sounds perfectly suited for a pod. You have failover, it doesn't persist to disk, and if you're already on kubernetes, it greatly simplifies deployments between environments.

Well, the transactor certainly writes to disk if you're using any fulltext attributes OR if you're using Valcache with the transactor. You may be able to configure a pod to have an ephemeral NVMe SSD for Valcache, I'm not sure.

port 4334 is exposed

root@lms-app-65bf78f8db-q5z95:/usr/src/app# telnet datomic-service 4334
Trying 10.100.252.126...
telnet: Unable to connect to remote host: Connection refused
root@lms-app-65bf78f8db-q5z95:/usr/src/app# telnet datomic-service 4330
Trying 10.100.252.126...

Is the peer in the cluster too?

peers communicate with storage independently (not via the transactor)

so datomic-service needs to resolve to postgres for peers, or else you can add alt-host= with an alternative hostname

yes the peer is in the cluster

host= is also the bind address for the transactor. What does datomic-service resolve to on the transactor? Did the transactor start up correctly?

the transactor started correctly

datomic-service resolves to itself on the transactor

Launching with Java options -server -Xms1g -Xmx1g -XX:+UseG1GC -XX:MaxGCPauseMillis=50
Starting datomic:sql://<DB-NAME>?jdbc:, you may need to change the user and password parameters to work with your jdbc driver ...
System started datomic:sql://<DB-NAME>?jdbc:, you may need to change the user and password parameters to work with your jdbc driver

is “resolves to itself” specifically what IP address? The exact same one: 10.100.252.126?

or maybe some internal address? or loopback, or…

telnet: can't connect to remote host (10.100.252.126): Connection refused

this is on the transactor

so yes, they're the same IP

I guess check with netstat

the transactor is just showing connections to postgres

the peer also shows a connection to postgres

this was all working before i attempted to put it in kubernetes. Same DB, which i wiped and reconnected.

when re-connected, it correctly updated the host url