Fork me on GitHub
#xtdb
<
2019-10-14
>
JanisOlex14:10:17

Hi, Crux ppl Could you tell me if there is anything in clojure/crux which might break coz I am trying to run 2 crux DB in one VM? Errors are random clojure compulations

JanisOlex14:10:49

common things seems to be - breakage on trying to compile something from taonesso

jonpither14:10:32

Hi @olekss.janis What configuration of Crux are you using? Kafka + KV?

JanisOlex14:10:12

breakes on i.db

JanisOlex14:10:00

but when it is alone -> it works, but when two such codes end up in the same VM, randomly breaks on taonesso compilation -> tandomly -> means really randomly, feels like real clash of some namespaces or overwriting something somewhere

jonpither14:10:17

Let me try and see if I can reproduce

JanisOlex14:10:14

to have more fun -> it all runs with quasar -javaagent, though I tried to put all crux and it's dependencies on quasar exclusion list

JanisOlex14:10:35

So not corda or quasar problem, pure this code gave exactly the same bug

JanisOlex14:10:51

Is there some cheap fix?

JanisOlex14:10:15

Running on windows

JanisOlex14:10:19

problem is I can't run sequentially, as I won't be able to control the node using crux startup during integration tests

jonpither14:10:00

could you elaborate on the latter point - how would you like to control the node?

JanisOlex14:10:36

I can't... that's the point, but seems crux has some issues or I am not smart enough to configure, but this simple code snipped prooves that 2 crux services started at the same VM breaks... I just checked the very code I posted with one instance, and it starts up no problem

jonpither14:10:00

OK, checking

JanisOlex14:10:00

so what is wrong with crux/clojure that it can't spin 2 nodes at the same time?

jonpither14:10:55

writing a test right now to verify.... hang tight

JanisOlex14:10:47

ok will be around - probably blinking in/out

JanisOlex15:10:38

@jonpither I rewrote just the code to run in sequence... and no problems... on first startup it took some time, probably due to compilation, and then very fast spinned up 2 versions without errors

jonpither15:10:39

Good to know. I've a failing test of sorts, but looks a bit different, am investigating https://github.com/juxt/crux/pull/340/commits/894732630c68e98d1df5facffd9ab9ef513e7292

jonpither15:10:47

note that test is in a branch, not master

JanisOlex15:10:28

in case if there is easy fix, ping me... now I am off

JanisOlex15:10:53

I am also free to bump up the dependency version in case if you release fix...

jonpither15:10:06

I also tried testing this against standalone, so I need to rethink my approach - will check something else

JanisOlex16:10:18

to be honest - didn't understood what you just told? Is there bug? Or I need to change something on my side? Or you will be rethinking approach how to implement something, or this bug is not so essential... ? 🙂

jonpither16:10:48

Sorry for being unclear - we can't find a bug after a few different tests now. When I said "rethinking our approach" - I meant coming up with ideas of tests to write, that would reproduce your error

Ivan Fedorov16:10:15

@olekss.janis could you please share your gradle.kts file?

jonpither16:10:22

defo my bad, for being unclear up there ^