Say, if I want to include another jar in the datomic transactor, do I just copy it into the resources directory? Specifically, I'm trying to get this logback appender working: https://github.com/logzio/logzio-logback-appender

@timgilbert Yes, the resources dir should be on the classpath if youre using the bin/transactor script to launch

Great, thanks

Hmm, actually it appears that doesn't work, the server gets started with java -server -cp lib/*:datomic-transactor-pro-0.9.5404.jar:samples/clj:bin:resources

...but just sticking a jar file inside resources doesn't work, I think the jar needs to be explicitly named

I tried sticking it in lib but the glob lib/* is never actually expanded

Making this change to bin/classpath seems to do the trick:

$ diff /tmp/cp bin/classpath
6c6
<   s="`echo lib/*`:`echo *transactor*.jar`"
---
>   s="lib/*:`echo *transactor*.jar`"

Hmm, on further inspection it looks like some kind of dependency problem with the appender, will poke around more.

Oct 24 19:05:19 ip-172-31-14-11 transactor[25414]: Reported exception:
Oct 24 19:05:19 ip-172-31-14-11 transactor[25414]: java.lang.NoSuchMethodError: ch.qos.logback.core.Context.getScheduledExecutorService()Ljava/util/concurrent/ScheduledExecutorService;
Oct 24 19:05:19 ip-172-31-14-11 transactor[25414]: #011at io.logz.logback.LogzioLogbackAppender.start(LogzioLogbackAppender.java:166)
Oct 24 19:05:19 ip-172-31-14-11 transactor[25414]: #011at ch.qos.logback.core.joran.action.AppenderAction.end(AppenderAction.java:96)

Ah, and I just realized that Java itself supports the foo/* glob syntax, so the shell has nothing to do with it

Ok, peering at the code more closely, it seems as though the problem is that the appender uses logback-classic 1.1.7, but the transactor itself uses 1.0.13

Does Datomic still have a 10 billion datom limitation?

I was looking into using it for a data warehousing application instead of data vault.

it’s not a hard limit @joshg, but an anticipation of the likely max size that a peer process can deal with

because peers need to hold onto all the roots of the index and still have space for actual data

theoretical limit. if you have biiiig instances, your limit is higher

afaik nothing in the code imposes a limit

i’m not sure about the entity id space limitations though, perhaps @stuarthalloway can share 🙂

Thank you for the clarification, that makes sense. So it’s possible to use Datomic with more datoms than 10 billion, but perhaps that wouldn’t be its best use case.