Re: Kubelet/node nice level
Got it. So now I just need to fix my scripts. Thanks for clarifying. -- Mateus Caruccio / Master of Puppets GetupCloud.com We make the infrastructure invisible Gartner Cool Vendor 2017 2018-07-01 14:18 GMT-03:00 Clayton Coleman : > That’s the one the installer lays down. Ansible has never used the one in > the RPMs (and the one in the RPMs is being removed in 3.10 to prevent > confusion). > > On Jul 1, 2018, at 10:03 AM, Mateus Caruccio com> wrote: > > Yep, I copy/paste from an old buffer. It's /bin/nice already. Same results. > > Anyway, I believe have found the reason. Looks like > /etc/systemd/system/multi-user.target.wants is linking to the wrong unit > file (or am I editing the wrong one?). > > # ls -l /etc/systemd/system/multi-user.target.wants/origin-node.service > lrwxrwxrwx. 1 root root 39 Jul 1 13:42 /etc/systemd/system/multi- > user.target.wants/origin-node.service -> > */etc/systemd/system/origin-node.service* > > # cat /etc/systemd/system/origin-node.service > [Unit] > Description=OpenShift Node > After=docker.service > After=chronyd.service > After=ntpd.service > Wants=openvswitch.service > After=ovsdb-server.service > After=ovs-vswitchd.service > Wants=docker.service > Documentation=https://github.com/openshift/origin > Wants=dnsmasq.service > After=dnsmasq.service > > [Service] > Type=notify > EnvironmentFile=/etc/sysconfig/origin-node > Environment=GOTRACEBACK=crash > ExecStartPre=/usr/bin/cp /etc/origin/node/node-dnsmasq.conf > /etc/dnsmasq.d/ > ExecStartPre=/usr/bin/dbus-send --system --dest=uk.org.thekelleys.dnsmasq > /uk/org/thekelleys/dnsmasq uk.org.thekelleys.SetDomainServers > array:string:/in-addr.arpa/127.0.0.1,/cluster.local/127.0.0.1 > ExecStopPost=/usr/bin/rm /etc/dnsmasq.d/node-dnsmasq.conf > ExecStopPost=/usr/bin/dbus-send --system --dest=uk.org.thekelleys.dnsmasq > /uk/org/thekelleys/dnsmasq uk.org.thekelleys.SetDomainServers > array:string: > ExecStart=/usr/bin/openshift start node --config=${CONFIG_FILE} $OPTIONS > LimitNOFILE=65536 > LimitCORE=infinity > WorkingDirectory=/var/lib/origin/ > SyslogIdentifier=origin-node > Restart=always > RestartSec=5s > TimeoutStartSec=300 > OOMScoreAdjust=-999 > > [Install] > WantedBy=multi-user.target > > > After remove it, the link points to where I expect being the right unit > file: > > > # rm -f /etc/systemd/system/origin-node.service > > # systemctl disable origin-node > Removed symlink /etc/systemd/system/multi-user.target.wants/origin-node. > service. > > # systemctl enable origin-node > Created symlink from /etc/systemd/system/multi- > user.target.wants/origin-node.service to /usr/lib/systemd/system/ > origin-node.service. > > *# systemctl restart origin-node* > > # ps ax -o pid,nice,comm|grep openshift > 4994 0 openshift > 5036 -5 openshift > > > The question now is: where /etc/systemd/system/origin-node.service comes > from? > > -- > Mateus Caruccio / Master of Puppets > GetupCloud.com > We make the infrastructure invisible > Gartner Cool Vendor 2017 > > 2018-07-01 8:06 GMT-03:00 Tobias Florek : > >> Hi! >> >> > ExecStart=nice -n -5 /usr/bin/openshift start node [...] >> >> That won't work. You need the full path to the executable in systemd >> units. >> >> Cheers, >> Tobias Florek >> >> ___ >> dev mailing list >> dev@lists.openshift.redhat.com >> http://lists.openshift.redhat.com/openshiftmm/listinfo/dev >> >> > ___ > dev mailing list > dev@lists.openshift.redhat.com > http://lists.openshift.redhat.com/openshiftmm/listinfo/dev > > ___ dev mailing list dev@lists.openshift.redhat.com http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
Re: Kubelet/node nice level
That’s the one the installer lays down. Ansible has never used the one in the RPMs (and the one in the RPMs is being removed in 3.10 to prevent confusion). On Jul 1, 2018, at 10:03 AM, Mateus Caruccio wrote: Yep, I copy/paste from an old buffer. It's /bin/nice already. Same results. Anyway, I believe have found the reason. Looks like /etc/systemd/system/multi-user.target.wants is linking to the wrong unit file (or am I editing the wrong one?). # ls -l /etc/systemd/system/multi-user.target.wants/origin-node.service lrwxrwxrwx. 1 root root 39 Jul 1 13:42 /etc/systemd/system/multi-user.target.wants/origin-node.service -> */etc/systemd/system/origin-node.service* # cat /etc/systemd/system/origin-node.service [Unit] Description=OpenShift Node After=docker.service After=chronyd.service After=ntpd.service Wants=openvswitch.service After=ovsdb-server.service After=ovs-vswitchd.service Wants=docker.service Documentation=https://github.com/openshift/origin Wants=dnsmasq.service After=dnsmasq.service [Service] Type=notify EnvironmentFile=/etc/sysconfig/origin-node Environment=GOTRACEBACK=crash ExecStartPre=/usr/bin/cp /etc/origin/node/node-dnsmasq.conf /etc/dnsmasq.d/ ExecStartPre=/usr/bin/dbus-send --system --dest=uk.org.thekelleys.dnsmasq /uk/org/thekelleys/dnsmasq uk.org.thekelleys.SetDomainServers array:string:/in-addr.arpa/127.0.0.1,/cluster.local/127.0.0.1 ExecStopPost=/usr/bin/rm /etc/dnsmasq.d/node-dnsmasq.conf ExecStopPost=/usr/bin/dbus-send --system --dest=uk.org.thekelleys.dnsmasq /uk/org/thekelleys/dnsmasq uk.org.thekelleys.SetDomainServers array:string: ExecStart=/usr/bin/openshift start node --config=${CONFIG_FILE} $OPTIONS LimitNOFILE=65536 LimitCORE=infinity WorkingDirectory=/var/lib/origin/ SyslogIdentifier=origin-node Restart=always RestartSec=5s TimeoutStartSec=300 OOMScoreAdjust=-999 [Install] WantedBy=multi-user.target After remove it, the link points to where I expect being the right unit file: # rm -f /etc/systemd/system/origin-node.service # systemctl disable origin-node Removed symlink /etc/systemd/system/multi-user.target.wants/origin-node.service. # systemctl enable origin-node Created symlink from /etc/systemd/system/multi-user.target.wants/origin-node.service to /usr/lib/systemd/system/origin-node.service. *# systemctl restart origin-node* # ps ax -o pid,nice,comm|grep openshift 4994 0 openshift 5036 -5 openshift The question now is: where /etc/systemd/system/origin-node.service comes from? -- Mateus Caruccio / Master of Puppets GetupCloud.com We make the infrastructure invisible Gartner Cool Vendor 2017 2018-07-01 8:06 GMT-03:00 Tobias Florek : > Hi! > > > ExecStart=nice -n -5 /usr/bin/openshift start node [...] > > That won't work. You need the full path to the executable in systemd > units. > > Cheers, > Tobias Florek > > ___ > dev mailing list > dev@lists.openshift.redhat.com > http://lists.openshift.redhat.com/openshiftmm/listinfo/dev > > ___ dev mailing list dev@lists.openshift.redhat.com http://lists.openshift.redhat.com/openshiftmm/listinfo/dev ___ dev mailing list dev@lists.openshift.redhat.com http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
Re: Kubelet/node nice level
Yep, I copy/paste from an old buffer. It's /bin/nice already. Same results. Anyway, I believe have found the reason. Looks like /etc/systemd/system/multi-user.target.wants is linking to the wrong unit file (or am I editing the wrong one?). # ls -l /etc/systemd/system/multi-user.target.wants/origin-node.service lrwxrwxrwx. 1 root root 39 Jul 1 13:42 /etc/systemd/system/multi-user.target.wants/origin-node.service -> */etc/systemd/system/origin-node.service* # cat /etc/systemd/system/origin-node.service [Unit] Description=OpenShift Node After=docker.service After=chronyd.service After=ntpd.service Wants=openvswitch.service After=ovsdb-server.service After=ovs-vswitchd.service Wants=docker.service Documentation=https://github.com/openshift/origin Wants=dnsmasq.service After=dnsmasq.service [Service] Type=notify EnvironmentFile=/etc/sysconfig/origin-node Environment=GOTRACEBACK=crash ExecStartPre=/usr/bin/cp /etc/origin/node/node-dnsmasq.conf /etc/dnsmasq.d/ ExecStartPre=/usr/bin/dbus-send --system --dest=uk.org.thekelleys.dnsmasq /uk/org/thekelleys/dnsmasq uk.org.thekelleys.SetDomainServers array:string:/in-addr.arpa/127.0.0.1,/cluster.local/127.0.0.1 ExecStopPost=/usr/bin/rm /etc/dnsmasq.d/node-dnsmasq.conf ExecStopPost=/usr/bin/dbus-send --system --dest=uk.org.thekelleys.dnsmasq /uk/org/thekelleys/dnsmasq uk.org.thekelleys.SetDomainServers array:string: ExecStart=/usr/bin/openshift start node --config=${CONFIG_FILE} $OPTIONS LimitNOFILE=65536 LimitCORE=infinity WorkingDirectory=/var/lib/origin/ SyslogIdentifier=origin-node Restart=always RestartSec=5s TimeoutStartSec=300 OOMScoreAdjust=-999 [Install] WantedBy=multi-user.target After remove it, the link points to where I expect being the right unit file: # rm -f /etc/systemd/system/origin-node.service # systemctl disable origin-node Removed symlink /etc/systemd/system/multi-user.target.wants/origin-node.service. # systemctl enable origin-node Created symlink from /etc/systemd/system/multi-user.target.wants/origin-node.service to /usr/lib/systemd/system/origin-node.service. *# systemctl restart origin-node* # ps ax -o pid,nice,comm|grep openshift 4994 0 openshift 5036 -5 openshift The question now is: where /etc/systemd/system/origin-node.service comes from? -- Mateus Caruccio / Master of Puppets GetupCloud.com We make the infrastructure invisible Gartner Cool Vendor 2017 2018-07-01 8:06 GMT-03:00 Tobias Florek : > Hi! > > > ExecStart=nice -n -5 /usr/bin/openshift start node [...] > > That won't work. You need the full path to the executable in systemd > units. > > Cheers, > Tobias Florek > > ___ > dev mailing list > dev@lists.openshift.redhat.com > http://lists.openshift.redhat.com/openshiftmm/listinfo/dev > > ___ dev mailing list dev@lists.openshift.redhat.com http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
Re: Kubelet/node nice level
Hi! > ExecStart=nice -n -5 /usr/bin/openshift start node [...] That won't work. You need the full path to the executable in systemd units. Cheers, Tobias Florek signature.asc Description: signature ___ dev mailing list dev@lists.openshift.redhat.com http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
Re: Kubelet/node nice level
Nice doesn't appears on origin-node.service. However I managed to sucessfully set it on both docker.service and crond.service. Also I've tried to call it using the nice command with no success: ExecStart=nice -n -5 /usr/bin/openshift start node --config=${CONFIG_FILE} $OPTIONS [root@ip-10-0-53-142 centos]# systemctl daemon-reload [root@ip-10-0-53-142 centos]# systemctl stop origin-node.service [root@ip-10-0-53-142 centos]# systemctl start origin-node.service [root@ip-10-0-53-142 centos]# ps afx -o pid,nice,comm|grep openshift 1744 0 openshift 1758 0 openshift 87650 0 openshift -- Mateus Caruccio / Master of Puppets GetupCloud.com We make the infrastructure invisible Gartner Cool Vendor 2017 2018-06-30 17:01 GMT-03:00 Clayton Coleman : > Maybe double check that systemd sees the Nice parameter with systemctl cat > origin-node > > On Jun 30, 2018, at 3:10 PM, Mateus Caruccio com> wrote: > > [centos@ip-10-0-53-142 ~]$ oc version > oc v3.9.0+ba7faec-1 > kubernetes v1.9.1+a0ce1bc657 > features: Basic-Auth GSSAPI Kerberos SPNEGO > > Server https://api.engine.caruccio.com > openshift v3.9.0+ba7faec-1 > kubernetes v1.9.1+a0ce1bc657 > > -- > > [centos@ip-10-0-53-142 ~]$ sudo grep -v '^#' /etc/sysconfig/origin-node > /etc/sysconfig/origin-node:OPTIONS=--loglevel=1 > /etc/sysconfig/origin-node:CONFIG_FILE=/etc/origin/node/node-config.yaml > /etc/sysconfig/origin-node: > /etc/sysconfig/origin-node:IMAGE_VERSION=v3.9.0 > /etc/sysconfig/origin-node:AWS_ACCESS_KEY_ID=[REDACTED] > /etc/sysconfig/origin-node:AWS_SECRET_ACCESS_KEY=[REDACTED] > > -- > > [centos@ip-10-0-53-142 ~]$ sudo cat /etc/origin/node/node-config.yaml > allowDisabledDocker: false > apiVersion: v1 > dnsBindAddress: 127.0.0.1:53 > dnsRecursiveResolvConf: /etc/origin/node/resolv.conf > dnsDomain: cluster.local > dnsIP: 10.0.53.142 > dockerConfig: > execHandlerName: "" > iptablesSyncPeriod: "30s" > imageConfig: > format: openshift/origin-${component}:${version} > latest: False > kind: NodeConfig > kubeletArguments: > cloud-config: > - /etc/origin/cloudprovider/aws.conf > cloud-provider: > - aws > image-gc-high-threshold: > - '80' > image-gc-low-threshold: > - '50' > image-pull-progress-deadline: > - 20m > max-pods: > - '200' > maximum-dead-containers: > - '10' > maximum-dead-containers-per-container: > - '1' > minimum-container-ttl-duration: > - 30s > node-labels: > - region=primary > - role=master > - zone=default > - server_name=mateus-master-0 > pods-per-core: > - '20' > masterClientConnectionOverrides: > acceptContentTypes: application/vnd.kubernetes.protobuf,application/json > contentType: application/vnd.kubernetes.protobuf > burst: 200 > qps: 100 > masterKubeConfig: system:node:ip-10-0-53-142.us-west-2.compute.internal. > kubeconfig > networkPluginName: redhat/openshift-ovs-multitenant > # networkConfig struct introduced in origin 1.0.6 and OSE 3.0.2 which > # deprecates networkPluginName above. The two should match. > networkConfig: >mtu: 8951 >networkPluginName: redhat/openshift-ovs-multitenant > nodeName: ip-10-0-53-142.us-west-2.compute.internal > podManifestConfig: > servingInfo: > bindAddress: 0.0.0.0:10250 > certFile: server.crt > clientCA: ca.crt > keyFile: server.key > minTLSVersion: VersionTLS12 > volumeDirectory: /var/lib/origin/openshift.local.volumes > proxyArguments: > proxy-mode: > - iptables > volumeConfig: > localQuota: > perFSGroup: > > -- > > [centos@ip-10-0-53-142 ~]$ sudo cat /usr/lib/systemd/system/ > origin-node.service > [Unit] > Description=Origin Node > After=docker.service > Wants=docker.service > Documentation=https://github.com/openshift/origin > > [Service] > Type=notify > EnvironmentFile=/etc/sysconfig/origin-node > Environment=GOTRACEBACK=crash > ExecStart=/usr/bin/openshift start node --config=${CONFIG_FILE} $OPTIONS > LimitNOFILE=65536 > LimitCORE=infinity > WorkingDirectory=/var/lib/origin/ > SyslogIdentifier=origin-node > Restart=always > RestartSec=5s > OOMScoreAdjust=-999 > Nice = -5 > > [Install] > WantedBy=multi-user.target > > -- > > This is the ansible task I'm using: > > https://github.com/caruccio/getup-engine-installer/commit/ > 49c28e4cc350856e11b8160f7a315e5fdda0dcce > > --- > - name: Set origin-node niceness > ini_file: > path: /usr/lib/systemd/system/origin-node.service > section: Service > option: Nice > value: -5 > backup: yes > tags: > - post-install > > > -- > Mateus Caruccio / Master of Puppets > GetupCloud.com > We make the infrastructure invisible > Gartner Cool Vendor 2017 > > 2018-06-30 15:03 GMT-03:00 Clayton Coleman : > >> Which version of openshift and what are your node start settings? >> >> On Jun 29, 2018, at 11:10 PM, Mateus Caruccio < >> mateus.caruc...@getupcloud.com> wrote: >> >> Hi. I'm trying to run openshift kubelet with nice
Re: Kubelet/node nice level
Maybe double check that systemd sees the Nice parameter with systemctl cat origin-node On Jun 30, 2018, at 3:10 PM, Mateus Caruccio wrote: [centos@ip-10-0-53-142 ~]$ oc version oc v3.9.0+ba7faec-1 kubernetes v1.9.1+a0ce1bc657 features: Basic-Auth GSSAPI Kerberos SPNEGO Server https://api.engine.caruccio.com openshift v3.9.0+ba7faec-1 kubernetes v1.9.1+a0ce1bc657 -- [centos@ip-10-0-53-142 ~]$ sudo grep -v '^#' /etc/sysconfig/origin-node /etc/sysconfig/origin-node:OPTIONS=--loglevel=1 /etc/sysconfig/origin-node:CONFIG_FILE=/etc/origin/node/node-config.yaml /etc/sysconfig/origin-node: /etc/sysconfig/origin-node:IMAGE_VERSION=v3.9.0 /etc/sysconfig/origin-node:AWS_ACCESS_KEY_ID=[REDACTED] /etc/sysconfig/origin-node:AWS_SECRET_ACCESS_KEY=[REDACTED] -- [centos@ip-10-0-53-142 ~]$ sudo cat /etc/origin/node/node-config.yaml allowDisabledDocker: false apiVersion: v1 dnsBindAddress: 127.0.0.1:53 dnsRecursiveResolvConf: /etc/origin/node/resolv.conf dnsDomain: cluster.local dnsIP: 10.0.53.142 dockerConfig: execHandlerName: "" iptablesSyncPeriod: "30s" imageConfig: format: openshift/origin-${component}:${version} latest: False kind: NodeConfig kubeletArguments: cloud-config: - /etc/origin/cloudprovider/aws.conf cloud-provider: - aws image-gc-high-threshold: - '80' image-gc-low-threshold: - '50' image-pull-progress-deadline: - 20m max-pods: - '200' maximum-dead-containers: - '10' maximum-dead-containers-per-container: - '1' minimum-container-ttl-duration: - 30s node-labels: - region=primary - role=master - zone=default - server_name=mateus-master-0 pods-per-core: - '20' masterClientConnectionOverrides: acceptContentTypes: application/vnd.kubernetes.protobuf,application/json contentType: application/vnd.kubernetes.protobuf burst: 200 qps: 100 masterKubeConfig: system:node:ip-10-0-53-142.us -west-2.compute.internal.kubeconfig networkPluginName: redhat/openshift-ovs-multitenant # networkConfig struct introduced in origin 1.0.6 and OSE 3.0.2 which # deprecates networkPluginName above. The two should match. networkConfig: mtu: 8951 networkPluginName: redhat/openshift-ovs-multitenant nodeName: ip-10-0-53-142.us-west-2.compute.internal podManifestConfig: servingInfo: bindAddress: 0.0.0.0:10250 certFile: server.crt clientCA: ca.crt keyFile: server.key minTLSVersion: VersionTLS12 volumeDirectory: /var/lib/origin/openshift.local.volumes proxyArguments: proxy-mode: - iptables volumeConfig: localQuota: perFSGroup: -- [centos@ip-10-0-53-142 ~]$ sudo cat /usr/lib/systemd/system/origin-node.service [Unit] Description=Origin Node After=docker.service Wants=docker.service Documentation=https://github.com/openshift/origin [Service] Type=notify EnvironmentFile=/etc/sysconfig/origin-node Environment=GOTRACEBACK=crash ExecStart=/usr/bin/openshift start node --config=${CONFIG_FILE} $OPTIONS LimitNOFILE=65536 LimitCORE=infinity WorkingDirectory=/var/lib/origin/ SyslogIdentifier=origin-node Restart=always RestartSec=5s OOMScoreAdjust=-999 Nice = -5 [Install] WantedBy=multi-user.target -- This is the ansible task I'm using: https://github.com/caruccio/getup-engine-installer/commit/49c28e4cc350856e11b8160f7a315e5fdda0dcce --- - name: Set origin-node niceness ini_file: path: /usr/lib/systemd/system/origin-node.service section: Service option: Nice value: -5 backup: yes tags: - post-install -- Mateus Caruccio / Master of Puppets GetupCloud.com We make the infrastructure invisible Gartner Cool Vendor 2017 2018-06-30 15:03 GMT-03:00 Clayton Coleman : > Which version of openshift and what are your node start settings? > > On Jun 29, 2018, at 11:10 PM, Mateus Caruccio com> wrote: > > Hi. I'm trying to run openshift kubelet with nice set o -5 but with no > success. > > I've noticed that kubelet is started using syscall.Exec[1], which calls > execve. The man page of execve[2] states that the new process shall > inherited nice value from caller. > > After adding `Nice=-5` to origin-node unit and reloading both daemon and > unit, openshift node process still runs with nice=0. > > What am I missing? > > [1]: https://github.com/openshift/origin/blob/ > 83ac5ae6a7d635ae67b1be438d85c339500fd65b/pkg/cmd/server/ > start/start_node.go#L433 > [2]: https://linux.die.net/man/3/execve > > -- > Mateus Caruccio / Master of Puppets > GetupCloud.com > We make the infrastructure invisible > Gartner Cool Vendor 2017 > > ___ > dev mailing list > dev@lists.openshift.redhat.com > http://lists.openshift.redhat.com/openshiftmm/listinfo/dev > > ___ dev mailing list dev@lists.openshift.redhat.com http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
Re: Kubelet/node nice level
[centos@ip-10-0-53-142 ~]$ oc version oc v3.9.0+ba7faec-1 kubernetes v1.9.1+a0ce1bc657 features: Basic-Auth GSSAPI Kerberos SPNEGO Server https://api.engine.caruccio.com openshift v3.9.0+ba7faec-1 kubernetes v1.9.1+a0ce1bc657 -- [centos@ip-10-0-53-142 ~]$ sudo grep -v '^#' /etc/sysconfig/origin-node /etc/sysconfig/origin-node:OPTIONS=--loglevel=1 /etc/sysconfig/origin-node:CONFIG_FILE=/etc/origin/node/node-config.yaml /etc/sysconfig/origin-node: /etc/sysconfig/origin-node:IMAGE_VERSION=v3.9.0 /etc/sysconfig/origin-node:AWS_ACCESS_KEY_ID=[REDACTED] /etc/sysconfig/origin-node:AWS_SECRET_ACCESS_KEY=[REDACTED] -- [centos@ip-10-0-53-142 ~]$ sudo cat /etc/origin/node/node-config.yaml allowDisabledDocker: false apiVersion: v1 dnsBindAddress: 127.0.0.1:53 dnsRecursiveResolvConf: /etc/origin/node/resolv.conf dnsDomain: cluster.local dnsIP: 10.0.53.142 dockerConfig: execHandlerName: "" iptablesSyncPeriod: "30s" imageConfig: format: openshift/origin-${component}:${version} latest: False kind: NodeConfig kubeletArguments: cloud-config: - /etc/origin/cloudprovider/aws.conf cloud-provider: - aws image-gc-high-threshold: - '80' image-gc-low-threshold: - '50' image-pull-progress-deadline: - 20m max-pods: - '200' maximum-dead-containers: - '10' maximum-dead-containers-per-container: - '1' minimum-container-ttl-duration: - 30s node-labels: - region=primary - role=master - zone=default - server_name=mateus-master-0 pods-per-core: - '20' masterClientConnectionOverrides: acceptContentTypes: application/vnd.kubernetes.protobuf,application/json contentType: application/vnd.kubernetes.protobuf burst: 200 qps: 100 masterKubeConfig: system:node:ip-10-0-53-142.us-west-2.compute.internal.kubeconfig networkPluginName: redhat/openshift-ovs-multitenant # networkConfig struct introduced in origin 1.0.6 and OSE 3.0.2 which # deprecates networkPluginName above. The two should match. networkConfig: mtu: 8951 networkPluginName: redhat/openshift-ovs-multitenant nodeName: ip-10-0-53-142.us-west-2.compute.internal podManifestConfig: servingInfo: bindAddress: 0.0.0.0:10250 certFile: server.crt clientCA: ca.crt keyFile: server.key minTLSVersion: VersionTLS12 volumeDirectory: /var/lib/origin/openshift.local.volumes proxyArguments: proxy-mode: - iptables volumeConfig: localQuota: perFSGroup: -- [centos@ip-10-0-53-142 ~]$ sudo cat /usr/lib/systemd/system/origin-node.service [Unit] Description=Origin Node After=docker.service Wants=docker.service Documentation=https://github.com/openshift/origin [Service] Type=notify EnvironmentFile=/etc/sysconfig/origin-node Environment=GOTRACEBACK=crash ExecStart=/usr/bin/openshift start node --config=${CONFIG_FILE} $OPTIONS LimitNOFILE=65536 LimitCORE=infinity WorkingDirectory=/var/lib/origin/ SyslogIdentifier=origin-node Restart=always RestartSec=5s OOMScoreAdjust=-999 Nice = -5 [Install] WantedBy=multi-user.target -- This is the ansible task I'm using: https://github.com/caruccio/getup-engine-installer/commit/49c28e4cc350856e11b8160f7a315e5fdda0dcce --- - name: Set origin-node niceness ini_file: path: /usr/lib/systemd/system/origin-node.service section: Service option: Nice value: -5 backup: yes tags: - post-install -- Mateus Caruccio / Master of Puppets GetupCloud.com We make the infrastructure invisible Gartner Cool Vendor 2017 2018-06-30 15:03 GMT-03:00 Clayton Coleman : > Which version of openshift and what are your node start settings? > > On Jun 29, 2018, at 11:10 PM, Mateus Caruccio com> wrote: > > Hi. I'm trying to run openshift kubelet with nice set o -5 but with no > success. > > I've noticed that kubelet is started using syscall.Exec[1], which calls > execve. The man page of execve[2] states that the new process shall > inherited nice value from caller. > > After adding `Nice=-5` to origin-node unit and reloading both daemon and > unit, openshift node process still runs with nice=0. > > What am I missing? > > [1]: https://github.com/openshift/origin/blob/ > 83ac5ae6a7d635ae67b1be438d85c339500fd65b/pkg/cmd/server/ > start/start_node.go#L433 > [2]: https://linux.die.net/man/3/execve > > -- > Mateus Caruccio / Master of Puppets > GetupCloud.com > We make the infrastructure invisible > Gartner Cool Vendor 2017 > > ___ > dev mailing list > dev@lists.openshift.redhat.com > http://lists.openshift.redhat.com/openshiftmm/listinfo/dev > > ___ dev mailing list dev@lists.openshift.redhat.com http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
Kubelet/node nice level
Hi. I'm trying to run openshift kubelet with nice set o -5 but with no success. I've noticed that kubelet is started using syscall.Exec[1], which calls execve. The man page of execve[2] states that the new process shall inherited nice value from caller. After adding `Nice=-5` to origin-node unit and reloading both daemon and unit, openshift node process still runs with nice=0. What am I missing? [1]: https://github.com/openshift/origin/blob/83ac5ae6a7d635ae67b1be438d85c339500fd65b/pkg/cmd/server/start/start_node.go#L433 [2]: https://linux.die.net/man/3/execve -- Mateus Caruccio / Master of Puppets GetupCloud.com We make the infrastructure invisible Gartner Cool Vendor 2017 ___ dev mailing list dev@lists.openshift.redhat.com http://lists.openshift.redhat.com/openshiftmm/listinfo/dev