Binary for x86_64-linux Broken

[loneaggie]

I can bundle but not use the pre-built binary for x86_64 Linux. This is on Ubuntu 24.04.03 LTS. I get the same result with versions 3.6.0 and 3.7.0 of the SDK.

The error dumps look like this:

ms [crit] [2003,2003] *** Fatal error encountered during exception handling (rev="c2439a404fa6b276862fe81150b1326bf98e6164", compiler="GNU 11.4.0", system="Linux-6.8.0-1031-azure", date="2025-09-29T19:34:06") ***
0ms [crit] [2003,2003] Caught unhandled std::exception-derived exception. what(): std::bad_cast
0ms [crit] [2003,2003] Call stack:
<backtrace not supported on this platform>
terminate called after throwing an instance of 'std::bad_cast'
what():  std::bad_cast

[avsej]

Thank you @loneaggie, I will check and it will be fixed in 3.7.1

[avsej]

@loneaggie what is the Ruby version that you are using? I have just ubuntu:24.04 image from dockerhub, and stock version from Ubuntu works just fine.

root@4479d10d3934:/# ruby -v
ruby 3.2.3 (2024-01-18 revision 52bb2ac0a6) [x86_64-linux-gnu]

Could you give more info about your environment? It will help me to reproduce and fix the problem.

What script are you using? Something simple like this?

require "couchbase"
Couchbase::Cluster.connect("couchbase://localhost", "Administrator", "password")

[loneaggie]

@avsej hopefully this helps. Please let me know if there is anything else I can provide.

The ruby version is:

ruby -v
ruby 3.3.8 (2025-04-09 revision b200bad6cd) [x86_64-linux]

Our connection setup looks something like this:

couch_options = Options::Cluster.new(enable_metrics: false, enable_tracing: false)
couch_options.authenticate('username', 'password')
Cluster.connect("couchbase://localhost", couch_options)

It is Ubuntu 24.04.3 LTS (Noble Numbat).

[avsej]

@loneaggie, I have just pulled this Ubuntu, but it has different version of ruby. How do you install yours?

# grep VERSION /etc/os-release  
VERSION_ID="24.04"
VERSION="24.04.3 LTS (Noble Numbat)"
VERSION_CODENAME=noble

# ruby -v
ruby 3.2.3 (2024-01-18 revision 52bb2ac0a6) [x86_64-linux-gnu]

[loneaggie]

@avsej It is essentially installed from source via postmodern/ruby-install. (https://github.com/postmodern/ruby-install)

[avsej]

Hm, still something else causing this crash. Here is my reproduction:

• ./Dockerfile to build image with ruby and couchbase SDK.

FROM ubuntu:24.04

ENV DEBIAN_FRONTEND=noninteractive

RUN apt-get update && \
    apt-get install -y \
    curl \
    git \
    build-essential \
    libssl-dev \
    libreadline-dev \
    zlib1g-dev \
    libyaml-dev \
    libsqlite3-dev \
    libxml2-dev \
    libxslt1-dev \
    libncurses5-dev \
    libffi-dev \
    libgdbm-dev \
    libgmp-dev \
    && rm -rf /var/lib/apt/lists/*

RUN curl -L https://github.com/postmodern/ruby-install/archive/refs/heads/master.tar.gz | tar xz && \
    cd ruby-install-master && \
    make install && \
    cd .. && \
    rm -rf ruby-install-master

RUN ruby-install ruby 3.3.8 --system

ENV PATH="/usr/local/lib/ruby/gems/3.3.0/bin:/usr/local/bin:$PATH"

RUN echo "export RUBY_VERSION=3.3.8" >> /etc/environment

RUN gem update --system && \
    gem install couchbase -v 3.7.0

RUN mkdir -p /app

WORKDIR /app

COPY connect_couchbase.rb .

RUN chmod +x connect_couchbase.rb

CMD ["ruby", "connect_couchbase.rb"]

• ./connect_couchbase.rb simple script that just connects and writes sample document into "default" bucket.

#!/usr/bin/env ruby
# frozen_string_literal: true

require "couchbase"
require "json"
require "time"

host = ENV["COUCHBASE_HOST"] || "localhost"
username = ENV["COUCHBASE_USERNAME"] || "Administrator"
password = ENV["COUCHBASE_PASSWORD"] || "password"
bucket_name = ENV["COUCHBASE_BUCKET"] || "default"

connection_string = "couchbase://#{host}"

begin
  options = Couchbase::Options::Cluster.new(enable_metrics: false, enable_tracing: false)
  options.authenticate(username, password)
  cluster = Couchbase::Cluster.connect(connection_string, options)

  puts "Connected to Couchbase cluster at #{connection_string}"

  bucket = cluster.bucket(bucket_name)
  puts "Using bucket: #{bucket_name}"

  collection = bucket.default_collection

  document_id = "sample_doc_#{Time.now.to_i}"
  result = collection.upsert(document_id, {"timestamp" => Time.now.iso8601})
  puts "Created document with #{document_id} with CAS: #{result.cas}"
  retrieved_doc = collection.get(document_id)
  puts "Retrieved document #{document_id} with CAS: #{retrieved_doc.cas}"
rescue Couchbase::Error::AuthenticationFailure => e
  puts "Authentication failed: #{e.message}"
  exit 1
rescue Couchbase::Error::BucketNotFound => e
  puts "Bucket not found: #{e.message}"
  exit 1
rescue Couchbase::Error::Timeout => e
  puts "Connection timeout: #{e.message}"
  exit 1
rescue StandardError => e
  puts "Unexpected error: #{e.class} - #{e.message}"
  puts e.backtrace.join("\n")
  exit 1
ensure
  cluster&.disconnect
  puts "Disconnected from Couchbase cluster"
end

• ./docker-compose.yml to start Couchbase single-node (latest version), wait until it will create the bucket and then start ruby container with the script.

services:
  server:
    image: couchbase:latest
    container_name: server
    ports:
      - "8091:8091"
      - "11210:11210"
      - "8092:8092"
      - "8093:8093"
      - "8094:8094"
      - "11211:11211"
    volumes:
      - couchbase_data:/opt/couchbase/var
    healthcheck:
      test: ["CMD", "curl", "-f", "-u", "Administrator:password", "http://localhost:8091/pools/default/buckets/default"]
      interval: 1s
      timeout: 60s
      retries: 60
      start_period: 10s
    command: >
      bash -xe -c "
        if [ ! -f /opt/couchbase/var/.couchbase_initialized ]; then
          /entrypoint.sh couchbase-server &
          PID=$!
          sleep 10
          couchbase-cli cluster-init -c localhost:8091 --cluster-username=Administrator --cluster-password=password --cluster-ramsize=512 --services=data,index,query
          couchbase-cli bucket-create -c localhost:8091 --username=Administrator --password=password --bucket=default --bucket-type=couchbase --bucket-ramsize=256
          touch /opt/couchbase/var/.couchbase_initialized
          wait $PID
        else
          /entrypoint.sh couchbase-server
        fi
      "

  ruby:
    build: .
    container_name: ruby
    tty: true
    depends_on:
      server:
        condition: service_healthy
    environment:
      - COUCHBASE_HOST=server
      - COUCHBASE_USERNAME=Administrator
      - COUCHBASE_PASSWORD=password
      - COUCHBASE_BUCKET=default
    entrypoint: ["sh", "-c", "ruby /app/connect_couchbase.rb"]

volumes:
  couchbase_data:

With all that setup, I run docker-compose, and everything just works.

$ docker-compose up
...
...
Container server Healthy 
ruby    | [2026-01-27 21:41:45.989]   56ms [warn] [7,12] DNS SRV query returned 0 records for "server", assuming that cluster is listening this address
ruby    | Connected to Couchbase cluster at couchbase://server
ruby    | [2026-01-27 21:41:46.098]  109ms [warn] [7,12] [f7714e-b060-c542-1311-bb2d302a5ec36e/default] unable to find connected session with GCCCP support, retry in 2500ms
ruby    | Using bucket: default
ruby    | Created document with sample_doc_1769550106 with CAS: 1769550106099318784
ruby    | Retrieved document sample_doc_1769550106 with CAS: 1769550106099318784
ruby    | Disconnected from Couchbase cluster
ruby exited with code 0

@loneaggie, do you have any ideas on what else might be missing? I really would like to get this issue fixed. Maybe the library was installed and then the system was upgraded, or something like that?