Bad gateway. Help needed

Support Developer support
1

Hello!

I’m getting a bad gateway error that I can’t solve. On the Opal R log I can see that my function is running, however when the return line is interpreted, a 502 error occurs.

A reproducible example:

require('DSI')
require('DSOpal')
library(dsBaseClient)
library(dsOmicsClient)

builder <- newDSLoginBuilder()
builder$append(server = "study1", url = "https://opal-demo.obiba.org", 
               user = "dsuser", password = "password", 
               resource = "RSRC.tcga_liver")

logindata <- builder$build()

conns <- datashield.login(logins = logindata, assign = TRUE, 
                          symbol = "res")

datashield.assign.expr(conns = conns, symbol = "rse", expr = quote(as.resource.object(res)))

ans.gender <- ds.limma(model =  ~ gdc_cases.demographic.gender, 
                       Set = "rse", type.data = "RNAseq", 
                       sva = FALSE)
deseq_ds = ds.DESeq2(model = ~ gdc_cases.demographic.gender, set = "rse")

On this example (when I run it) the ds.limma returns a table while the ds.DESeq2 returns a 502 error. Both functions are called using the same structure, so I can’t understand why one works and the other does not. The code for them can be found at

Thanks for your time,

Xavier

2

Hi Xavier, in my experience this type of error happens when the server is overloaded. In my local machine using DESeq2 is a lot more intensive than limma. Limma runs OK, but the equivalent DESeq2 command on the same data causes a crash. The error is different though - I get a connection refused message

3

Those are unfortunate news, thanks for the response.

4

Hi Xavier,

It is time for you to get your own development server :slight_smile: opal-demo is not meant for heavy loads.

Best
Yannick

5

Yes, definitely I will have to look into getting a development server. Do you think it’s possible to develop some sort of check / error formatting that catches this types of crashes? I got really frustrated not understanding the source of my error.

6

The problem is that R eats all the memory and opal has no room to react/reply. One option is to have distinct servers, one for opal, another one for R.

Yannick

7

By the way, if the your DS aggregation task takes a long time to run, have a try to the dev version of DSI:

remotes::install_github("datashield/DSI")
remotes::install_github("datashield/DSOpal")
remotes::install_github("datashield/DSLite")

it might help.

Regards
Yannick