Error: curl_multi_socket_action: 8

[vdayanand]

To reproduce

using Downloads
function main()
           url = "https://julialang-s3.julialang.org/bin/linux/x64/0.6/julia-0.6.0-linux-x86_64.tar.gz"
           infos = [url, url]
           Downloads.download(url)
               @sync for info in infos
                   @async try
                       Downloads.download(info)
                   catch err
                       @warn "Failed to download binary" exception=(err, catch_backtrace())
                   end
               end

       end
main()

└ @ Downloads.Curl /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.6/Downloads/src/Curl/utils.jl:36
┌ Error: curl_multi_socket_action: 8
└ @ Downloads.Curl /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.6/Downloads/src/Curl/utils.jl:36
┌ Error: curl_multi_socket_action: 8
└ @ Downloads.Curl /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.6/Downloads/src/Curl/utils.jl:36
┌ Error: curl_multi_socket_action: 8
└ @ Downloads.Curl /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.6/Downloads/src/Curl/utils.jl:36
┌ Error: curl_multi_socket_action: 8
└ @ Downloads.Curl /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.6/Downloads/src/Curl/utils.jl:36
┌ Error: curl_multi_socket_action: 8

julia version info:

Julia Version 1.6.5
Commit 9058264a69 (2021-12-19 12:30 UTC)
Platform Info:
  OS: Linux (x86_64-pc-linux-gnu)
  CPU: Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-11.0.1 (ORCJIT, broadwell)
Environment:
  JULIA_GPG = 3673DF529D9049477F76B37566E3C7DC03D6E495
  JULIATEAM_OS = UBUNTU
  JULIAHUB_MODE = false
  JULIA_VERSION = 1.6.5
  JULIA_DEPOT_PATH = /juliateam/depot
  JULIA_PATH = /usr/local/julia
  JULIA_PKG_SERVER = http://juliateamwebsrvr:9999

[pfitzseb]

Dup of https://github.com/JuliaLang/Downloads.jl/issues/110. That issue's title isn't correct since it applies to downloads on multiple tasks as well.

[StefanKarpinski]

Yeah, known issue on 1.6 and the changes to make it threadsafe are too significant to backport (unless we decide that it's ok to backport significant features in stdlibs to the LTS version of Julia, which would certainly make my life easier).