容器进程request_wait_answer状态处理
用户存储被禁用后在保存容器镜像时发现部分进程卡住查看进程栈发现是读写FUSE文件系统时进程正在等待FUSE响应request_wait_answer宿主机上查看rootwlss01:/sys/fs/cgroup/kubepods.slice/kubepods-kubepods.slice/kubepods-kubepods-pod525dc8be_262b_4c8c_bd51_555b76c79ca5.slice/cri-containerd-d4c5ef5c6f4275946c70f046e1e48792c35274bbe862282e309018621b7429ff.scope# cat /proc/2669982/stack [0] request_wait_answer0x15f/0x2a0 [0] fuse_simple_request0x18d/0x2f0 [0] fuse_flush0x1d5/0x210 [0] filp_flush0x38/0x90 [0] filp_close0x14/0x30 [0] do_dup20x9c/0xf0 [0] ksys_dup30xa6/0x130 [0] __x64_sys_dup20x29/0xd0 [0] x64_sys_call0x23f7/0x25a0 [0] do_syscall_640x7f/0x180 [0] entry_SYSCALL_64_after_hwframe0x78/0x80容器内查看28142rootgpu1:/workspace# ps -eo pid,stat,wchan:32,cmd PID STAT WCHAN CMD 1 Ss - /pause 7 Ss do_wait bash /entrypoint.sh 52 Ss do_poll.constprop.0 sshd: /usr/sbin/sshd [listener] 0 of 10-100 startups 53 S ep_poll /usr/bin/python /usr/local/bin/jupyter-lab --ip0.0.0.0 --no-browser --allow-root 54 Ss - bash /workspace/memcheck.sh 2866 Ss do_wait /bin/bash 8578 T do_signal_stop top -c 25570 T do_signal_stop /bin/bash 1.sh 25707 T do_signal_stop sleep 30 28142 Ss request_wait_answer /bin/bash 35777 S - sleep 1 35778 R - ps -eo pid,stat,wchan:32,cmd目前的处理办法是在用户保存容器镜像时如果是因存储禁用导致的request_wait_answer先进行kill确保容器能够正常pause。
