Freeze on "terminate" a big recorging

[jujumo]

Describe the bug

I am running "run_image_slam" inside "stella_vslam-desktop" docker container.
The image sequence is made of 38k images.
Everything goes fine until I click "terminate".
Then, my system freeze.

It works fine on shorter image sequence.

To Reproduce

When running ... I get ""

shell session:

xhost +local: && docker run -it --rm \
    --gpus all --privileged -e DISPLAY=$DISPLAY -v /tmp/.X11-unix/:/tmp/.X11-unix:ro \
    --volume ./equi:/dataset:ro --volume ./equi_trajectory:/output \
    stella_vslam-desktop \
    -c "/stella_vslam_examples/build/run_image_slam \
        -v /output/orb_vocab.fbow \
        -d /dataset/ \
        -c /output/config.yaml \
        --frame-skip 3 --no-sleep --auto-term \
        -o /output/map.msg"

COMPLETE FREEZE

build options:

docker build -t stella_vslam-desktop -f Dockerfile.desktop . --build-arg NUM_THREADS=`expr $(nproc) - 1`

config file:

# Equirectangular model for RICOH THETA S

Camera:
  name: "insta360"
  setup: "monocular"
  model: "equirectangular"

  fps: 29.96
  cols: 1920
  rows: 960

  color_order: "RGB"

Preprocessing:
  min_size: 100
  mask_rectangles:
    - [0.0, 1.0, 0.0, 0.1]
    - [0.0, 1.0, 0.84, 1.0]
    - [0.0, 0.2, 0.7, 1.0]
    - [0.8, 1.0, 0.7, 1.0]

Feature:
  name: "default ORB feature extraction setting"
  scale_factor: 1.15
  num_levels: 12
  ini_fast_threshold: 10
  min_fast_threshold: 5
  max_num_keypoints: 2500

Mapping:
  backend: "g2o"
  baseline_dist_thr_ratio: 0.02
  redundant_obs_ratio_thr: 0.90
  num_covisibilities_for_landmark_generation: 20
  num_covisibilities_for_landmark_fusion: 20
  residual_deg_thr: 0.4

Tracking:
  backend: "g2o"

LoopDetector:
  backend: "g2o"
  enabled: true
  reject_by_graph_distance: true
  min_distance_on_graph: 50

GraphOptimizer:
  min_num_shared_lms: 200

GlobalOptimizer:
  thr_neighbor_keyframes: 100

System:
  map_format: "msgpack"
  num_grid_cols: 96
  num_grid_rows: 48

Expected behavior

save "map.msg"

Screenshots or videos

Environment

• Hardware: PC, ubuntu with nvidia GC
• OS: ubuntu 24.04
• Install procedure: docker
• dataset: custom 38K images