Skip to content

[NLP] Crash reported in the pytorch_inference process  #2571

Open
@davidkyle

Description

@davidkyle

The error message as logged in Elasticsearch is:

[2023-09-20T20:24:02,682][ERROR][o.e.x.m.i.d.DeploymentManager] [ml-ES8-elastic-qa025] [sentence-transformers__distiluse-base-multilingual-cased-v1] inference process crashed due to reason [[sentence-transformers__distiluse-base-multilingual-cased-v1] pytorch_inference/821 process stopped unexpectedly: Fatal error: 'The futex facility returned an unexpected error code.', version: 8.9.1 (build a285a437dd4bb2)
Fatal error: 'si_signo 11, si_code: 128, si_errno: 0, address: 0x7f680100d941, library: /lib/x86_64-linux-gnu/libc.so.6, base: 0x7f6800feb000, normalized address: 0x22941', version: 8.9.1 (build a285a437dd4bb2)
]

The crash was reported on this platform:

#uname -a
#224-Ubuntu SMP Mon Jun 19 13:30:12 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux
# cat /proc/cpuinfo | head -n 5
processor      : 0
vendor_id      : GenuineIntel
cpu family     : 6
model          : 85
model name     : Intel Xeon Processor (Skylake, IBRS)

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions