| Name |
Last commit
|
Last update |
|---|---|---|
| .. | ||
| systemd | ||
| README.md | ||
| deploy-docs.sh | ||
| deploy-prod.sh | ||
| deploy.sh | ||
| first-import.sh | ||
| gen-env.sh | ||
| health-watch.sh | ||
| pac-backup.sh |
09-05 在测试机用同一个 pac-service 镜像、256MB 上限的探针跑通了整条链, 两处原写法与实测不符: 1. 「越界 → SIGKILL(exit 137)」误导排查方向。 restart:always 下 docker inspect 的 State.ExitCode / OOMKilled 显示的是 **重启后那条命** —— 实测内核明确 CONSTRAINT_MEMCG 杀了 2 次,inspect 同时 报 ExitCode=0 OOMKilled=false。照原文案去查会得出"没被杀"的反结论。 判据改为 RestartCount + dmesg CONSTRAINT_MEMCG,告警正文里也一并改。 2. 补上实测数:rss=241MB 时 heapUsed 仅 4MB。 这是 mem_limit 不可替代的硬证据 —— 堆上限对堆外(Buffer/engine)零约束力, 也是"别把 max-old-space-size 和 mem_limit 划等号"的依据。 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
| Name |
Last commit
|
Last update |
|---|---|---|
| .. | ||
| systemd | Loading commit data... | |
| README.md | Loading commit data... | |
| deploy-docs.sh | Loading commit data... | |
| deploy-prod.sh | Loading commit data... | |
| deploy.sh | Loading commit data... | |
| first-import.sh | Loading commit data... | |
| gen-env.sh | Loading commit data... | |
| health-watch.sh | Loading commit data... | |
| pac-backup.sh | Loading commit data... |