Fork me on GitHub
#datomic
<
2017-04-10
>
cjmurphy10:04:57

@kwladyka (you asked a jetty clj-client deps question late last year, seems similar to this) or others - possible to solve this SO question? http://stackoverflow.com/questions/43291069/lein-ring-server-headless-fails-when-including-datomic-dependency#comment73705091_43291069

kwladyka10:04:45

@cjmurphy i think i asked… year ago 😉 I found solution.

kwladyka10:04:32

but not really remember… it was something about broken dependency i think..

kwladyka10:04:03

anyway i am using http-kit now instead of jetty

cjmurphy10:04:26

Seems similar - the asking person might come back here. I'm not really a datomic person and never had to solve a triangular dependency problem myself so going to have to give up. Seems strange there's no easy was to go back to a prior version of clj-client for instance.

kwladyka10:04:14

[com.datomic/datomic-free "0.9.5544"] [ring/ring-core "1.5.0"] [http-kit "2.2.0"] - it works for me

cjmurphy10:04:29

Thanks - I'll relay 🙂

kwladyka10:04:42

heh ok 🙂

dominicm11:04:26

I notice that d/datoms doesn't have any guarantees about laziness in the docstring, is that intentional?

isaac15:04:56

should datomic use datasource for sql schema?

djjolicoeur19:04:34

does anyone have experience running a higher than normal throughput transactor? We are currently running on the 4G set up from the docs and appear to be outgrowing it. wondering if there are tradeoffs associated with going above the 4G recommended settings and increasing the object-cache size?

marshall19:04:07

@djjolicoeur yes I’ve run higher heaps fairly consistently for load testing/etc What in particular indicates to you you’re outgrowing 4g?

djjolicoeur19:04:08

@marshall you mind if I DM you specifics?

pbostrom19:04:59

How do folks handle failover in a HA transactor configuration on AWS? I understand that the standby transactor is going take over if the primary transactor does not send a heartbeat. But how do you signal to kill the instance of the failed transactor? My first instinct is to just have something that checks port 4334 and if it's not listening just kill the instance, but I'm wondering if anyone has any other ideas

marshall19:04:18

@pbostrom The provided Datomic AMI launch script calls shutdown -h now when it detects that the transactor process terminates

marshall19:04:52

choose your favorite bash-fu method of process monitoring 🙂

marshall19:04:54

the problem with checking 4334 would be that it wouldn’t work for the standby transactor; i believe the port is only opened once the transactor becomes active

pbostrom20:04:49

makes sense, thanks @marshall. is the launch script available anywhere? I decided to build my own AMI

marshall20:04:47

i’m not sure if it’s publicly accessible or not it really doesn’t do much. basically downloads the bits, unzips them, and runs bin/transactor

csm22:04:51

one thing I’ve done is register a health callback with datomic, that just dumps the stats it gets to a file

csm22:04:56

I then have a simple web server that responds OK if that file has been written to within a timeout, so it works as an ELB health check