0

I am running Spark 2.4.4 using Yarn and interfacing using RSparkling and Sparklyr

As per these instructions I've

  1. Installed Sparklyr
  2. Called the library for Sparklyr
  3. Removed any prior installs of H2O
  4. Installed the latest version of H2O (rel-zorn)
  5. Installed rsparkling 3.36.0.3-1-2.4
  6. Called the library for rsparkling
  7. Specified my spark_config()
  8. Successfully made a connection to Spark using Yarn
  9. Ran h2oConf <- H2OConf()

When I try to make a H2O context using the h2oConf above I get the following error:

Error in h2o.init(strict_version_check = FALSE, https = https, insecure = insecure,  : 
  unused argument (cacert = conf$sslCACert())

I've tried multiple different versions of RSparkling and H2O and have been unsuccessful connecting.

Is there some obvious step that I'm missing? Any suggestions would be greatly appreciated.

Solution: Based on feedback from @Marek Novotny below I discovered that I had a reference to an old version of H2O in my namespace. Once I unloaded the package I was able to resolve this issue and moved on to the next

TheGoat
  • 2,587
  • 3
  • 25
  • 58

1 Answers1

1

It seems that your environment still contains old H2O R library. cacert is an valid parameter and it was introduced in H2O 3.26.0.6.

  • I've removed all references to old H2O installations using # REMOVE PRIOR INSTALLS OF H2O if ("package:h2o" %in% search()) { detach("package:h2o", unload=TRUE) } if ("h2o" %in% rownames(installed.packages())) { remove.packages("h2o") } – TheGoat Mar 25 '22 at 13:54
  • Is there something else I should be doing to flush out old versions? – TheGoat Mar 25 '22 at 13:54
  • What do you get when you execute: `packageVersion("h2o")`? – Marek Novotny Mar 25 '22 at 15:06
  • 1
    or `getNamespaceVersion("h2o")` – Marek Novotny Mar 25 '22 at 15:08
  • getNamespaceVersion("h2o") returned an old version of h20. I am in the process of restarting R and unloading the namespace to clear any old references. – TheGoat Mar 25 '22 at 15:31