UI issue with R plots

[lkahn]

Starting on May 6, our workspace in ap-southeast-2 stopped being able to display R plots correctly. This is specific to this workspace (we have 3), and specific to R plots, and is occurring for all users and clusters.

The figure itself can be resized using R code (it's not an issue of the plot dimensions) but the Databricks UI is only giving the plot a tiny sliver for display, so all we see is the top sliver of the figure.

Testing it further shows that other weird things are happening – displayed dataframes are sometimes overlaid over the figure display area, and sometimes figures don’t display at all (see attached).

How can I resolve this? 

[plankton]

I have the same issue.

https://community.databricks.com/t5/data-engineering/r-plots-not-rendering/m-p/157353#M54536

[Lu_Wang_ENB_DBX]

This looks like a workspace-side notebook UI regression. 

Recommendation

1. Open a Databricks support ticket and treat it as a Notebook UI / rendering regression for that workspace. Include:

   • Workspace ID
   • Region: ap-southeast-2
   • Start date: May 6
   • A notebook URL
   • Screenshots
   • A HAR capture
   • The fact that it affects all users and clusters, and only R plot output in that workspace

   Ask support to check for a workspace-scoped frontend/webapp rollout or flag regression in the notebook rendering path.

2. Use a workaround instead of inline R plot output until support fixes it. Databricks R notebooks normally capture plots as inline PNGs, and notebooks also support displayHTML() for custom HTML output.

library(ggplot2)

p <- ggplot(mtcars, aes(wt, mpg)) + geom_point()

dir.create("/dbfs/FileStore/tmp/rplots", recursive = TRUE, showWarnings = FALSE)
fname <- sprintf("plot-%s.png", as.integer(Sys.time()))
fullpath <- file.path("/dbfs/FileStore/tmp/rplots", fname)

ggsave(fullpath, plot = p, width = 10, height = 6, dpi = 150)

displayHTML(sprintf(
  "<img src='/files/tmp/rplots/%s' style='max-width:100%%; height:auto;'/>",
  fname
))

3. If you need an immediate operational workaround, run those R notebooks in one of your unaffected workspaces until the workspace UI issue is fixed.