容器运行报错线程创建失败

redhat容器内部执行程序报错如下

Starting program: /home/soft/HPCServer/./HPCServer
warning: Error disabling address space randomization: Operation not permitted
Missing separate debuginfos, use: dnf debuginfo-install glibc-2.34-105.oe2203sp1.x86_64
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/usr/lib64/libthread_db.so.1".
warning: Ignoring non-absolute filename: <libnsl.so.1>
Missing separate debuginfo for libnsl.so.1
Try: dnf --enablerepo='*debug*' install /usr/lib/debug/.build-id/96/fffd935288296af9a07b535c736e0e376ba55a.debug
warning: Ignoring non-absolute filename: <./libsasl2.so.3>
Missing separate debuginfo for ./libsasl2.so.3
Try: dnf --enablerepo='*debug*' install /usr/lib/debug/.build-id/57/d7f694ff1bff339da151ca41399c5c94a6933a.debug
hpr tls index{2}
hlog.async=>_sync=1
SyncLogHandler!
schina !!! HPR_CreateThread_Local HPR_Thread_Create error, return -1 3

schina !!! HPR_InitThread_Local HPR_CreateThread_Local error, return -1 2

schina !!! HPR_ThreadPool_Create HPR_InitThread_Local error, return -1 3

Program received signal SIGSEGV, Segmentation fault.
0x00007fc9839a1081 in __gnu_cxx::__exchange_and_add (__val=-1, __mem=0x3030303030307838) at /usr/include/c++/4.8/ext/atomicity.h:49
49 /usr/include/c++/4.8/ext/atomicity.h: No such file or directory.
Missing separate debuginfos, use: dnf debuginfo-install e2fsprogs-1.46.4-17.oe2203sp1.x86_64 keyutils-libs-1.6.3-4.oe2203sp1.x86_64 krb5-libs-1.19.2-6.oe2203sp1.x86_64 libgcc-10.3.1-20.x86_64 libselinux-3.3-2.oe2203sp1.x86_64 libstdc++-10.3.1-20.x86_64 libxcrypt-4.4.26-4.oe2203sp1.x86_64 pcre2-10.39-6.oe2203sp1.x86_64 zlib-1.2.11-22.oe2203sp1.x86_64

 

解决方法：
修改dockersrun命令

#!/bin/bash
docker stop hpcserver
docker rm hpcserver
docker run -d \
  --name hpcserver \
  --restart=unless-stopped \
  -p 0.0.0.0:30990:30990 \
  -p 0.0.0.0:31250:1250/udp \
  -e HPC_MSGHANDER_IP="10.19.170.33" \
  -e MY_POD_NAME=hpcserver \
  --cpu-period=100000 \
  --cpu-quota=100000 \
  --memory=1005Mi \
  --ulimit nproc=65535:65535 \
  --ulimit nofile=65535:65535 \
  --security-opt seccomp=unconfined \
  --cap-add=SYS_PTRACE \
  --mount type=bind,source=/home/ssd/AlgorithmLib,destination=/home/ssd/AlgorithmLib \
  --mount type=bind,source=/home/ssd/HMSProxy,destination=/home/ssd/HMSProxy \
  --mount type=bind,source=/home/ssd/log/HPCServer/,destination=/home/ssd/log/HPCServer/ \
  --mount type=bind,source=/opt/opsmgr/web/opsMgrAgent/data/custom/,destination=/opt/opsmgr/web/opsMgrAgent/data/custom/ \
  docker.hikcloud:30001/hpc-deepmind/hpcserver:V4.1.0-20260707

1. --ulimit nproc=65535:65535 （最重要）

• 增加容器内可创建的最大进程/线程数
• 解决 HPR_Thread_Create error 问题

2. --ulimit nofile=65535:65535

• 增加文件描述符限制
• 防止因文件描述符不足导致的问题

3. --security-opt seccomp=unconfined

• 禁用 seccomp 安全策略
• 允许程序使用 clone 系统调用创建线程
• 解决 errno = 1 (EPERM) 问题

4. --cap-add=SYS_PTRACE

• 添加 ptrace 系统调用权限
• 允许程序进行调试和跟踪