リモートサーバーで強化学習を試すとlibGL error: failed to load driver: swrastが起きます

Question

リモートサーバー、機械学習ともに初心者です。強化学習のコードをGoogle Cloud Engine（Ubuntu 10) で一から学習させようとしているのですが、worldmodels.data.sample_policyでサンプルを集めようとすると以下のエラーが起きます。調べましたが、内容の専門性が高すぎて理解できません。問題とできれば解決策を噛み砕いて説明していただけないでしょうか？

(python-project-3.6.12) server_name:~/world-models$ xvfb-run -a -s "-screen 0 1400x900x24 +ext
ension RANDR" -- python3 -m worldmodels.data.sample_policy --num_process 8 --policy random

Track generation: 1195..1498 -> 303-tiles track
Track generation: 1176..1474 -> 298-tiles track
Track generation: 1105..1385 -> 280-tiles track
Track generation: 1199..1503 -> 304-tiles track
Track generation: 1205..1517 -> 312-tiles track
Track generation: 1242..1557 -> 315-tiles track
Track generation: 1185..1495 -> 310-tiles track
Track generation: 1135..1423 -> 288-tiles track
libGL error: unable to load driver: swrast_dri.so
libGL error: failed to load driver: swrast
libGL error: unable to load driver: swrast_dri.so
libGL error: failed to load driver: swrast
libGL error: unable to load driver: swrast_dri.so
libGL error: failed to load driver: swrast
libGL error: unable to load driver: swrast_dri.so
libGL error: failed to load driver: swrast
libGL error: unable to load driver: swrast_dri.so
libGL error: failed to load driver: swrast
libGL error: unable to load driver: swrast_dri.so
libGL error: failed to load driver: swrast
libGL error: unable to load driver: swrast_dri.so
libGL error: failed to load driver: swrast
libGL error: unable to load driver: swrast_dri.so
libGL error: failed to load driver: swrast
Traceback (most recent call last):
  File "/home/server/.pyenv/versions/3.6.12/lib/python3.6/runpy.py", line 193, in _run_module_as_main
    "__main__", mod_spec)
  File "/home/server/.pyenv/versions/3.6.12/lib/python3.6/runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "/home/kumikonakajima_2021/world-models/worldmodels/data/sample_policy.py", line 209, in <module>
    range(num_process)
  File "/home/server/.pyenv/versions/python-project-3.6.12/lib/python3.6/site-packages/pathos/multiproce
ssing.py", line 137, in map
    return _pool.map(star(f), zip(*args)) # chunksize
　File "/home/server/.pyenv/versions/python-project-3.6.12/lib/python3.6/site-packages/multiprocess/pool
.py", line 266, in map
    return self._map_async(func, iterable, mapstar, chunksize).get()
  File "/home/server/.pyenv/versions/python-project-3.6.12/lib/python3.6/site-packages/multiprocess/pool
.py", line 644, in get
    raise self._value
multiprocess.pool.MaybeEncodingError: Error sending result: '<multiprocess.pool.ExceptionWithTraceback object at 0x7f
a5b01f75f8>'. Reason: 'NameError("name 'base' is not defined",)'

元のソースコード（worldmodels.data.sample_policy）には以下の変更を加えました（PicklingErrorを直すため。参照：https://stackoverflow.com/questions/8804830/python-multiprocessing-picklingerror-cant-pickle-type-function）

from multiprocessing import Pool

から

from pathos.multiprocessing import ProcessingPool as Pool

ライブラリ等
requirement.txt
gym pyvirtualdisplay
xvfb python-opengl ffmpeg
pathos

マシーンタイプ
e2-standard-2 (2 vCPUs, 8 GB memory)
CPU platform: AMD Rome

Comments

[tiitoi]

基本的にopenglを使うソフトはリモートサーバーでは動かないです
ローカルマシンで動かす必要があります