From bdcb06bfcf46fbfcf7533dafe68417b6997de258 Mon Sep 17 00:00:00 2001 From: roc Date: Sat, 14 Oct 2023 14:55:16 +0800 Subject: [PATCH] update at 2023-10-14 14:55:16 --- content/best-practices/ha/smooth-upgrade.md | 2 +- content/tencent/appendix/ingress-error-code.md | 2 +- content/troubleshooting/cases/network/dns-lookup-5s-delay.md | 2 +- content/troubleshooting/node/no-space-left-on-device.md | 4 ++-- 4 files changed, 5 insertions(+), 5 deletions(-) diff --git a/content/best-practices/ha/smooth-upgrade.md b/content/best-practices/ha/smooth-upgrade.md index 91cc852..254c9e0 100644 --- a/content/best-practices/ha/smooth-upgrade.md +++ b/content/best-practices/ha/smooth-upgrade.md @@ -40,4 +40,4 @@ command: ["/bin/bash", "-c", "sleep 10"] ``` -最后，业务本身也需要实现优雅终止，避免被销毁时中断业务，参考 [优雅终止最佳实践](../graceful-shutdown/index.html) +最后，业务本身也需要实现优雅终止，避免被销毁时中断业务，参考 [优雅终止最佳实践](../graceful-shutdown) diff --git a/content/tencent/appendix/ingress-error-code.md b/content/tencent/appendix/ingress-error-code.md index d5993e1..00698e1 100644 --- a/content/tencent/appendix/ingress-error-code.md +++ b/content/tencent/appendix/ingress-error-code.md @@ -415,7 +415,7 @@ kubernetes.io/ingress.https-rules: '[{"path":"/abc","backend":{"serviceName":"ng ## E4052 Ingress Host Invalid -故障原因: 根据K8S的限制，Ingress的Host需要满足正则表达式 "(\*|[a-z0-9]([-a-z0-9]*[a-z0-9])?)(\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)+" +故障原因: 根据K8S的限制，Ingress的Host需要满足正则表达式 `(\*|[a-z0-9]([-a-z0-9]*[a-z0-9])?)(\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)+` 处理办法: 默认情况下域名都是符合以上要求的。排除一下域名没有 “.”，域名包含特殊字符等情况就可以了。 diff --git a/content/troubleshooting/cases/network/dns-lookup-5s-delay.md b/content/troubleshooting/cases/network/dns-lookup-5s-delay.md index 76ddd53..833aa8e 100644 --- a/content/troubleshooting/cases/network/dns-lookup-5s-delay.md +++ b/content/troubleshooting/cases/network/dns-lookup-5s-delay.md @@ -23,7 +23,7 @@ timeout:n 根本原因是内核 conntrack 模块的 bug，netfilter 做 NAT 时可能发生资源竞争导致部分报文丢弃。 -Weave works的工程师 [Martynas Pumputis](martynas@weave.works) 对这个问题做了很详细的分析：[Racy conntrack and DNS lookup timeouts](https://www.weave.works/blog/racy-conntrack-and-dns-lookup-timeouts) +Weave works 的工程师 `Martynas Pumputis` 对这个问题做了很详细的分析：[Racy conntrack and DNS lookup timeouts](https://www.weave.works/blog/racy-conntrack-and-dns-lookup-timeouts) 相关结论： diff --git a/content/troubleshooting/node/no-space-left-on-device.md b/content/troubleshooting/node/no-space-left-on-device.md index 14b479b..3ada248 100644 --- a/content/troubleshooting/node/no-space-left-on-device.md +++ b/content/troubleshooting/node/no-space-left-on-device.md @@ -13,7 +13,7 @@ Jul 18 15:20:58 VM_16_16_centos kubelet[11519]: E0718 15:20:58.280275 11519 raw.go:140] Failed to watch directory "/sys/fs/cgroup/memory/kubepods": inotify_add_watch /sys/fs/cgroup/memory/kubepods/burstable/pod926b7ff4-7bff-11e8-945b-52540048533c/6e85761a30707b43ed874e0140f58839618285fc90717153b3cbe7f91629ef5a: no space left on device ``` -系统调用 `inotify_add_watch` 失败，提示 `no space left on device`， 这是因为系统上进程 watch 文件目录的总数超出了最大限制，可以修改内核参数调高限制，详细请参考本书 [处理实践: inotify watch 耗尽](../../../handle/runnig-out-of-inotify-watches/) +系统调用 `inotify_add_watch` 失败，提示 `no space left on device`， 这是因为系统上进程 watch 文件目录的总数超出了最大限制，可以修改内核参数调高限制，详细请参考本书 [inotify watch 耗尽](runnig-out-of-inotify-watches.md) ## cgroup 泄露 @@ -82,4 +82,4 @@ limits: memory: 512 # 没有单位默认为字节，太小，应带上单位 ``` -根因：可能是因为内存相关的 volume 都受 memory limit 限制 (projected volume, emptydir 等)。 \ No newline at end of file +根因：可能是因为内存相关的 volume 都受 memory limit 限制 (projected volume, emptydir 等)。