NPD v0.8.10 release

Changes: v0.8.9...v0.8.10

CHANGELOG:

• Add kube-proxy health checker on linux. (#575, @uthark)
• Fix a systemstatsmonitor nil pointer panic. (#587, @vteratipally)
• Remove aufs hung check from the kernel monitor. (#596, @lizhuqi)
• Fix a critical issue in NPD healthchecker on windows, that would consume 100% cpu. (#609, #612, @mcshooter)

Docker Image: k8s.gcr.io/node-problem-detector/node-problem-detector:v0.8.10

node-problem-detector-v0.8.10-linux_amd64.tar.gz:
SHA: c127a794bc152f1dc930c90efbab711fc7d2cd8d
MD5: a30d8aa66e5fbef534106ef8bfebf042

node-problem-detector-v0.8.10-linux_arm64.tar.gz:
SHA: dcf895136422c19a09166defab18f20ef59f03f0
MD5: 249981c6396553ffc23db77c5321b962

node-problem-detector-v0.8.10-windows_amd64.tar.gz:
SHA: 4713cf43c9f38c7769c4f9d5ec128ffed85bf80a
MD5: 78e36628ceaeffdd16db49bc6cb62060

NPD v0.8.9 release

Changes: v0.8.8...v0.8.9

CHANGELOG:

• Imporved Health Checker
  • Added loopbacktime to reduce time of journalctl call
  • Added health-ckecker deploy file
    (#539, @smileusd)
• Improve Windows support
  • Added Windows Depenfer monitor (#567, @mcshooter)
  • Add HCS empty layer error reporting (#585, @jeremyje)
  • Support building binaries for Windows separately (#574, @mcshooter)
• Added ARM support (#561, @pwschuurman)
• Updated base docker debian image to fix some CVE (#577, @teratipally)
• Log error from kube-api
  (#583, @pezzak)
• Refactor disk_collector
  (#578, @teratipally)
• Fix e2e-test flakes for Ext4 counter (#570, @pwschuurman)

Docker Image: k8s.gcr.io/node-problem-detector/node-problem-detector:v0.8.9

node-problem-detector-v0.8.9-linux_amd64.tar.gz:
SHA: 9e10d673691c0a3533daa87f243e354b78d94dfa
MD5: 4a4b7ac8881c4ea22a9380c82aa3bf08

node-problem-detector-v0.8.9-linux_arm64.tar.gz:
SHA: ff21de69ca52140066855601ee9bce77ca7f0f5c
MD5: 5e12bc08bae5de29bf5968b6521ce328

node-problem-detector-v0.8.9-windows_amd64.tar.gz:
SHA: e08b867da00d075775387cec0ac03ace0b1a01e0
MD5: 1aba6948d441920b4148376a07a1496c

NPD v0.8.8 release

Changes: v0.8.7...v0.8.8

CHANGELOG:

• Added windows support
  • Build NPD binary and include in the release tarball (#517, #545, @jeremyje)
  • Support containerd log monitor on windows (#517, #542, #550, @jeremyje @mcshooter)
  • Support running NPD as a windows service (#541, @jeremyje)
  • Support containerd health checker on windows (#544, @mcshooter)
  • Support kube-proxy health checker on windows (#552, @mcshooter)
  • Support basic system metrics on windows (#554, @jeremyje)
  • Support windows defender problem detection (#555, @mcshooter)
• Add taint tolerations to the NPD yaml (#551, @cmssczy)
• Fix the known-modules.json relative path in the system stats monitor (#557, @vteratipally)
• Fix health checker uptime timestamp parsing (#558, @Random-Liu)

Docker Image: k8s.gcr.io/node-problem-detector/node-problem-detector:v0.8.8

node-problem-detector-v0.8.8-linux_amd64.tar.gz:
SHA: cae22ddd87bf00655f5eae559ee36f4a7c5b70c8
MD5: 317afff93e5a89aaf48fff274fb4d2a1

node-problem-detector-v0.8.8-windows_amd64.tar.gz:
SHA: f3bc3ab76d6a4a326c89297910c38ff615a39a0c
MD5: 4c2886ef86a883182b2c590b5f9d5ade

NPD v0.8.7 release

Changes: v0.8.6...v0.8.7

CHANGELOG:

• Fix an issue that kubelet may be restarted by NPD health checker unexpectedly (fix: kubernetes/kubernetes#87615),
  Make log pattern check configurable in health checker (#536, @abansal4032)
• Add memory read error (#530, @goushicui)
• Add support for building NPD on MacOS (#529, @stmcginnis)
• Exit the process when there is a timeout in plugin daemon (#521, @ZYecho)
• Add support for containerd health check (#518, @Random-Liu)
• Add metric for per-cpu, per-stage timing (#516, @karangoel)

Docker Image: k8s.gcr.io/node-problem-detector/node-problem-detector:v0.8.7

SHA: 739be1e17a391b5a06f0014bf8dbaf0d03c518d6
MD5: 41bfe82e6ce5183594e43ec609848a81

NPD v0.8.6 release

Changes: v0.8.5...v0.8.6

CHANGELOG:

• Add metric to retrieve stats such as procs_running, procs_blocked (https://github.com/kubernetes/node-probelm-detector/pulls/513, @karan)
• Add metric to retrieve network stats such as
  (https://github.com/kubernetes/node-probelm-detector/pulls/512, @karan)
• Introduce windows build of Node Problem Detector (https://github.com/kubernetes/node-probelm-detector/pulls/502, @jeremyje)
• Add a metric to retrieve guest os features such as unknown modules, ktd, kernel integrity (https://github.com/kubernetes/node-probelm-detector/pulls/492, @vteratipally)
• Fix custom plugin command timeout when the command spawns long running child process (https://github.com/kubernetes/node-probelm-detector/pulls/456, @zhangyue)
• Print result's message when status unknown (https://github.com/kubernetes/node-probelm-detector/pulls/455, @zhangyue)

Docker Image: k8s.gcr.io/node-problem-detector/node-problem-detector:v0.8.6

SHA: 403bc7d956c293fd16a9abe676bc11b3a9ec6830
MD5: ae1f8a80acff9a561569e7a07e1aff62

NPD v0.8.5 release

Changes: v0.8.4...v0.8.5

CHANGELOG:

• Fix the deployment yaml to prevent NPD from scheduling onto windows nodes. (#430, @wawa0210)
• Catching hung task with pattern like "tasks airflow scheduler: *".
  (#475, @vteratipally)
• Fix memory unit for /proc/meminfo metrics. (#478, neoseele@)
• Fix OOMKilling detection for new linux kernel v5.1+. (#481, tosi3k@)
• Add problem detection for buffer I/O error (#488, @vteratipally)
• Add CPU load averages metrics support. (#484, @karan)
• Add kubelet apiserver connection check in health checker (#489, @abansal4032)
• Avoid duplicating the disk bytes used metrics based on fstype and mount types (#497, @vteratipally)

Docker Image: k8s.gcr.io/node-problem-detector/node-problem-detector:v0.8.5

SHA: d830be1efc1f360aab1a7025da993ccc435836e2
MD5: 4ef926eca827f1593b22c45be99ec8fc

NPD v0.8.4 release

Changes: v0.8.3...v0.8.4

CHANGELOG:

• Add FSType and MountOption as labels to the metric disk_usage_bytes. (#452, @vteratipally)
• Add DockerContainerStartupFailure event in docker-monitor.json to detect docker issue docker/for-linux#647. (#453, @vteratipally)
• Reduce log spam generated by the custom plugin monitor (#458, #459, @abansal4032)

Docker Image: k8s.gcr.io/node-problem-detector/node-problem-detector:v0.8.4

SHA: 61eca11bdbd9db53d1a10fdbc5dcdbf8976947b3
MD5: e1d80ba826ae890936cb1c8270d0bf22

NPD v0.8.3 release

Changes: v0.8.2...v0.8.3

CHANGELOG:

• Include health-checker binary in the docker image (#433, @bengadbois)
• Set --enable-repair=true in the default config for docker and kubelet health checker (#443, @abansal4032)
• Fix a race condition that services may be killed periodically when --enable-repair=true, and systemd service restart time equals the health check period. (#444, @abansal4032)
• Only generate status update log when status changes in custom plugin (#441, @abansal4032 )
• Limit the size of custom plugin output to 4kb, extra output will be drained and discarded. And capture stderr output of the custom plugin and include it in the NPD log. (#442, @abansal4032 )

Docker Image: k8s.gcr.io/node-problem-detector/node-problem-detector:v0.8.3

SHA: 3216b8693932d77d0d33fc0010f2e9e100f73592
MD5: 7473734e4486f038d7e54046a9a178f1

NPD v0.8.2 release

Changes: v0.8.1...v0.8.2

CHANGELOG:

• Add a --event-namespace flag to make event namespace configurable (#418, @muff1nman)
• Fix a bug that lsblk is missing from the NPD container image (#421, @majst01)
• Add rehl support in OS version (#424, @stpabhi)
• Add health-checker as a custom plugin. The health-checker can be used to monitor healthiness of kubelet, docker and CRI container runtimes, e.g. containerd, cri-o, and restart them if they are not healthy if enable-repair is turned on. (#426, @abansal4032)

Docker Image: k8s.gcr.io/node-problem-detector:v0.8.2

SHA: 4c3fead23a98129278112d51b0cf897219ab56f1
MD5: 73b8a58b33c8fbcb15f94089bb6c8374

NPD v0.8.1 release

Changes: v0.8.0...v0.8.1

CHANGELOG:

• Support host_uptime metrics for CentOS (#379, @CuZn13)
• Fix build with build tag ENABLE_JOURNALD=0. (#390, @xueweiz )
• Improve network_problem.sh to support nf_conntrack and report error when conntrack table is 90% full. (#369, @arekkusu)
• Fix the first 0 value metrics reported for disk_avg_queue_len. (#402, @gmemcc)
• Fix a few metric units for disk metrics and the calculation for disk_avg_queue_len (#404, @xueweiz )
• Collect a lot more useful CPU/disk/memory metrics (#410, @xueweiz )

Docker Image: k8s.gcr.io/node-problem-detector:v0.8.1

SHA: 0f25126c5c85bc66e5a066e92e12958c93459799
MD5: d8633787dd1dd2142baa2787908165b6