目前的堆栈看起来不足以支撑做出判断; 如果线上有开core dump的话,可以用gdb palo_be core_dump文件看看堆栈 后者看下be.out是否进程怪盗时的堆栈
刘波 <270309...@qq.com.invalid> 于2022年3月4日周五 12:25写道: > 尊敬的开发者,您好: > 我们的doris > be节点经常会挂掉其中几个,经分析资源情况正常,具体信息可参见附件,日志层面未发现是OOM,未能定位出异常,请求协助,具体信息如下: > *时间: 2022-03-04 10:54:43* > doris报错*:detailMessage = tablet 45664199 has few replicas: 1, > alive backends: [10004]* > 环境信息:华为云cetnos 8 64位,16核64G > > 系统日志: less /var/log/message(*非OOM*) > > Mar 4 10:54:38 narwal-doris-be-0004 systemd[1]: Started Process Core Dump > (PID 1881819/UID 0). > > Mar 4 10:54:41 narwal-doris-be-0004 systemd-coredump[1881820]: Core file > was truncated to 2147483648 bytes. > > Mar 4 10:55:00 narwal-doris-be-0004 systemd-coredump[1881820]: Process > 2800630 (palo_be) of user 0 dumped core.#012#012Stack trace of thread > 1877601:#012#0 0x00000000039480a2 memcpy > > (/mnt/be/lib/palo_be)#012#012Stack trace of thread 2800630:#012#0 > 0x00007fccb330efc8 n/a (n/a) > > Mar 4 10:55:01 narwal-doris-be-0004 systemd[1]: > systemd-coredump@1-1881819-0.service: Succeeded. > > Mar 4 10:55:03 narwal-doris-be-0004 systemd[1]: session-12.scope: > Succeeded. > > > dump信息:*coredumpctl info 2800630* > > PID: 2800630 (palo_be) > UID: 0 (root) > GID: 0 (root) > Signal: 11 (SEGV) > Timestamp: Fri 2022-03-04 10:54:38 CST (1h 10min ago) > Command Line: /mnt/be/lib/palo_be > Executable: /mnt/be/lib/palo_be > Control Group: / > Slice: -.slice > Boot ID: ca3ef395d7c547d2aecb1c251097066f > Machine ID: 501f93b5c19d4ca38db845c29176e3c5 > Hostname: narwal-doris-be-0004 > Storage: > /var/lib/systemd/coredump/core.palo_be.0.ca3ef395d7c547d2aecb1c251097066f.2800630.1646362478000000.lz4 > (truncated) > Message: Process 2800630 (palo_be) of user 0 dumped core. > > Stack trace of thread 1877601: > #0 0x00000000039480a2 memcpy (/mnt/be/lib/palo_be) > > Stack trace of thread 2800630: > #0 0x00007fccb330efc8 n/a (n/a) > > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@doris.apache.org > For additional commands, e-mail: dev-h...@doris.apache.org -- 王博 Wang Bo