Welina mai i kā mākou papahana Kubernetes Quick Start. He kolamu maʻamau kēia me nā nīnau hoihoi loa i loaʻa iā mākou ma ka pūnaewele a ma kā mākou aʻo ʻana. Pane akamai Kubernetes.
ʻO Daniel Polenchik ka loea o kēia lā (). Hana ʻo Daniel ma ke ʻano he kumu aʻo a hoʻomohala polokalamu ma .
Inā makemake ʻoe e pane ʻia kāu nīnau ma ka pou aʻe, ole .
Ua hala i nā memo mua? .
Pehea e hoʻohui ai i nā pūʻulu Kubernetes ma nā kikowaena ʻikepili like ʻole?
Ka-pōkole: , a ke paipai nei au e heluhelu e pili ana и .
ʻO ka pinepine, hana hou ʻia ka ʻōnaehana a hoʻohele ʻia ma nā wahi like ʻole, ʻoi aku hoʻi i nā kaiapuni i hoʻomalu ʻia.
Inā ʻaʻole loaʻa kekahi ʻāpana, hoʻohuli ʻia ke kaʻa i kahi ʻē aʻe e pale aku i nā pilikia.
Me nā Kubernetes, hiki iā ʻoe ke hoʻohana i kahi hoʻolālā like a puʻunaue i nā haʻahaʻa hana ma nā wahi like ʻole.
Hiki iā ʻoe ke loaʻa i hoʻokahi a ʻoi aʻe paha nā puʻupuʻu no kēlā me kēia hui, ʻāina, kaiapuni, a i ʻole hui pū ʻana o kēia mau mea.
Hiki ke hoʻokipa ʻia kāu mau puʻupuʻu ma nā ao like ʻole a ma ka hale.
Akā, pehea ʻoe e hoʻolālā ai i nā ʻōnaehana no ia pālahalaha ʻāina?
Pono ʻoe e hana i hoʻokahi puʻupuʻu nui no nā kaiapuni ao ma luna o kahi pūnaewele hoʻokahi?
A i ʻole he nui nā puʻupuʻu liʻiliʻi a loaʻa i kahi ala e hoʻokele a hoʻonohonoho pū iā lākou?
Hoʻokahi pūʻulu alakaʻi
ʻAʻole maʻalahi ka hana ʻana i hoʻokahi puʻupuʻu ma kahi pūnaewele hoʻokahi.
E noʻonoʻo ʻoe he ulia pōpilikia, nalowale ka pilina ma waena o nā ʻāpana cluster.
Inā loaʻa iā ʻoe hoʻokahi kikowaena haku, ʻaʻole hiki i ka hapalua o nā kumuwaiwai ke loaʻa nā kauoha hou no ka mea ʻaʻole hiki iā lākou ke hoʻopili i ka haku.
A i ka manawa like loaʻa iā ʻoe nā papa routing kahiko (kube-proxy ʻAʻole hiki ke hoʻoiho i nā mea hou) a ʻaʻohe pods hou (ʻaʻole hiki i kubelet ke noi i nā mea hou).
ʻOi aku ka maikaʻi, inā ʻaʻole ʻike ʻo Kubernetes i kahi node, hōʻailona ia he keiki makua ʻole a hāʻawi i nā pods i nalowale i nā node e kū nei.
ʻO ka hopena, loaʻa iā ʻoe ka nui o nā pods.
Inā ʻoe e hana i hoʻokahi haku kikowaena no kēlā me kēia wahi, e loaʻa nā pilikia me ka consensus algorithm i ka waihona etcd. (kokoke. ed. - ʻOiaʻiʻo, ʻaʻole pono e loaʻa ka waihona etcd ma nā kikowaena master. Hiki ke holo ma kahi hui o nā kikowaena ma ka ʻāina like. ʻOiaʻiʻo, i ka manawa like e loaʻa ai kahi helu o ka hāʻule ʻole o ka hui. Akā wikiwiki.)
etcd hoʻohana e kūkākūkā i ka waiwai ma mua o ke kākau ʻana i ka diski.
ʻO ia hoʻi, ʻo ka hapa nui o nā hihia e pono e hōʻea i ka ʻaelike ma mua o ka hiki ke kākau ʻia ka mokuʻāina iā etcd.
Inā piʻi nui ka latency ma waena o nā manawa etcd, e like me ka hihia me ʻekolu etcd manawa ma nā wahi like ʻole, lōʻihi ka manawa e kūkākūkā ai i kahi waiwai a kākau iā ia i ka disk.
Hōʻike ʻia kēia i nā mea hoʻokele Kubernetes.
Pono ka luna hoʻomalu i ka manawa hou e aʻo ai e pili ana i ka hoʻololi a kākau i ka pane i ka waihona.
A no ka mea, ʻaʻole hoʻokahi mea hoʻoponopono, akā he nui. ka hopena o ke kaulahao a hoʻomaka ka hui holoʻokoʻa e hana mālie loa.
etcd ua maʻalahi loa ia .
ʻAʻole i kēia manawa nā hiʻohiʻona maikaʻi o kahi pūnaewele nui no kahi hui hoʻokahi.
ʻO ke kumu, ke hoʻāʻo nei ke kaiāulu hoʻomohala a me ka hui SIG-cluster e noʻonoʻo pehea e hoʻolalelale ai i nā puʻupuʻu e like me ke ʻano o Kubernetes orchestrates ipu.
Koho 1: hui hui pū me kubefed
Pane kūhelu mai SIG-cluster - .
No ka manawa mua, ua hoʻāʻo mākou e hoʻokele i ka hōʻiliʻili o nā puʻupuʻu ma ke ʻano he mea hoʻokahi me ka hoʻohana ʻana i ka mea hana kube federation.
Maikaʻi ka hoʻomaka ʻana, akā i ka hopena ʻaʻole i kaulana ʻo kube federation no ka mea ʻaʻole ia i kākoʻo i nā kumuwaiwai āpau.
Kākoʻo ia i nā hoʻopuka a me nā lawelawe i hui pū ʻia, akā ʻaʻole ʻo StatefulSets, no ka laʻana.
Eia kekahi, ua hoʻouna ʻia ka hoʻonohonoho federation ma ke ʻano o nā annotations a ʻaʻole maʻalahi.
E noʻonoʻo pehea e hiki ai iā ʻoe ke wehewehe i ka hoʻokaʻawale ʻana no kēlā me kēia puʻupuʻu i kahi hui me ka hoʻohana ʻana i nā annotation wale nō.
He haunaele loa.
Ua hana nui ʻo SIG-cluster ma hope o kubefed v1 a hoʻoholo e hoʻokokoke i ka pilikia mai kahi kihi ʻē aʻe.
Ma kahi o nā annotation, ua hoʻoholo lākou e hoʻokuʻu i kahi mea hoʻoponopono i hoʻokomo ʻia ma nā pūʻulu. Hiki ke hoʻopilikino ʻia me ka hoʻohana ʻana i ka Custom Resource Definitions (CRDs).
No kēlā me kēia kumuwaiwai e lilo i ʻāpana o ka hui, loaʻa iā ʻoe kahi wehewehe CRD maʻamau me ʻekolu mau ʻāpana:
- wehewehe maʻamau o kahi kumuwaiwai, no ka laʻana hoʻolaha;
- pauku
placement, kahi āu e wehewehe ai pehea e puʻunaue ʻia ai ka waiwai i ka hui; - pauku
override, kahi e hiki ai iā ʻoe ke hoʻopau i ke kaumaha a me nā ʻāpana mai kahi kau.
Eia kekahi laʻana o ka lawe ʻana i hui pū ʻia me ka waiho ʻana a me ka hoʻopau ʻana i nā ʻāpana.
apiVersion: types.federation.k8s.io/v1alpha1
kind: FederatedDeployment
metadata:
name: test-deployment
namespace: test-namespace
spec:
template:
metadata:
labels:
app: nginx
spec:
replicas: 3
selector:
matchLabels:
app: nginx
template:
metadata:
labels:
app: nginx
spec:
containers:
- image: nginx
name: nginx
placement:
clusterNames:
- cluster2
- cluster1
overrides:
- clusterName: cluster2
clusterOverrides:
- path: spec.replicas
value: 5E like me kāu e ʻike ai, hāʻawi ʻia ka lako ma nā ʻāpana ʻelua: cluster1 и cluster2.
Hāʻawi ka hui mua i ʻekolu kope, a ʻo ka lua i hoʻonohonoho ʻia i 5.
Inā makemake ʻoe i ka mana hou aʻe i ka helu o nā replicas, hāʻawi ʻo kubefed2 i kahi mea ReplicaSchedulingPreference hou kahi e hiki ke kaupaona ʻia nā kope:
apiVersion: scheduling.federation.k8s.io/v1alpha1
kind: ReplicaSchedulingPreference
metadata:
name: test-deployment
namespace: test-ns
spec:
targetKind: FederatedDeployment
totalReplicas: 9
clusters:
A:
weight: 1
B:
weight: 2ʻAʻole mākaukau loa ka ʻōnaehana CRD a me API, a ke hoʻomaka nei ka hana ikaika ma ka waihona o ka papahana.
E makaʻala i ka kubefed2, akā e hoʻomanaʻo ʻaʻole kūpono ia no ka hana ʻana.
E aʻo hou e pili ana i kubefed2 mai ma ka blog e pili ana i nā Kubernetes a ma .
Koho 2: hoʻohui i nā pūʻulu ma ke ʻano Booking.com
ʻAʻole i hana nā mea hoʻomohala o Booking.com ma kubefed v2, akā ua hele mai lākou me Shipper - kahi mea hoʻohana no ka lawe ʻana ma nā ʻāpana he nui, ma nā wahi a me nā ao.
ʻano like me kubefed2.
Hāʻawi nā mea hana ʻelua iā ʻoe e hana i kāu hoʻolālā hoʻolālā multi-cluster (ʻo ia nā pūʻulu i hoʻohana ʻia a me ka nui o nā replicas).
aka, ʻO ka pahuhopu o Shipper e hōʻemi i ka pilikia o nā hewa i ka wā o ka lawe ʻana.
Ma Shipper, hiki iā ʻoe ke wehewehe i ke ʻano o nā ʻanuʻu e wehewehe i ka mahele o nā replicas ma waena o ka hoʻolaha mua a me kēia manawa a me ka nui o nā kaʻa e hiki mai ana.
Ke hoʻokuʻu ʻoe i kahi kumuwaiwai i kahi puʻupuʻu, ʻo ka mea hoʻokele Shipper e hoʻololi i kēlā hoʻololi ma nā hui āpau i hui pū ʻia.
Eia kekahi, palena loa ʻo Shipper.
No kekahi laʻana, ʻae ia i nā palapala hoʻokele ma ke ʻano he hoʻokomo ʻaʻole kākoʻo i nā kumuwaiwai vanilla.
Ma nā ʻōlelo maʻamau, hana ʻo Shipper e like me kēia.
Ma kahi o ka hoʻouna maʻamau, pono ʻoe e hana i kahi kumuwaiwai noi e loaʻa kahi pakuhi Helm:
apiVersion: shipper.booking.com/v1alpha1
kind: Application
metadata:
name: super-server
spec:
revisionHistoryLimit: 3
template:
chart:
name: nginx
repoUrl: https://storage.googleapis.com/shipper-demo
version: 0.0.1
clusterRequirements:
regions:
- name: local
strategy:
steps:
- capacity:
contender: 1
incumbent: 100
name: staging
traffic:
contender: 0
incumbent: 100
- capacity:
contender: 100
incumbent: 0
name: full on
traffic:
contender: 100
incumbent: 0
values:
replicaCount: 3He koho maikaʻi ʻo Shipper no ka hoʻokele ʻana i nā puʻupuʻu lehulehu, akā ʻo kona pilina pili me Helm wale nō ke hele i ke ala.
Pehea inā e hoʻololi mākou a pau mai Helm i ai ole ia, ?
E ʻike hou aʻe e pili ana iā Shipper a me kāna ʻatikala ma .
Inā makemake ʻoe e ʻeli i loko o ke code, .
Koho 3: hui pū ʻia "magic".
Hana ʻo Kubefed v2 a me Shipper me ka hui pūʻulu, e hāʻawi ana i nā kumuwaiwai hou i nā pūʻulu ma o ka wehewehe kumu waiwai maʻamau.
Akā pehea inā ʻaʻole ʻoe makemake e kākau hou i nā hoʻopuka āpau, StatefulSets, DaemonSets, etc. e hui?
Pehea e hoʻokomo ai i kahi pūʻulu i loko o kahi hui me ka ʻole e hoʻololi i ka YAML?
, e pili ana i ka hoʻonohonoho ʻana i nā haʻawina hana ma nā pūʻulu.
Akā ma kahi o ka hele ʻana mai me kahi ala hou e launa pū ai me ka puʻupuʻu a hoʻopili i nā kumuwaiwai i nā wehewehe maʻamau, ua hoʻokomo ʻia ka multi-cluster-scheduler i loko o ke ola holoʻokoʻa Kubernetes maʻamau a kāpae i nā kelepona āpau e hana i nā pods.
Hoʻololi koke ʻia kēlā me kēia pod i hana ʻia me kahi dummy.
hoʻohana multi-cluster-scheduler e hoʻopaʻa i ke kelepona a hana i kahi pahu dummy hana ʻole.
E hele ana ka pod mua ma kahi pōʻai hoʻolālā ʻē aʻe, ma hope o ke koho balota ʻana i ka hui holoʻokoʻa, hoʻoholo ʻia kahi hoʻoholo.
ʻO ka hope, hāʻawi ʻia ka pod i ka puʻupuʻu pahuhopu.
ʻO ka hopena, loaʻa iā ʻoe kahi pod keu i hana ʻole, lawe wale i kahi.
ʻO ka pōmaikaʻi, ʻaʻole pono ʻoe e kākau i nā kumuwaiwai hou e hoʻohui i nā lako.
ʻO kēlā me kēia kumuwaiwai e hana i kahi pod ua mākaukau e hoʻohui ʻia.
He mea hoihoi kēia, no ka mea, ua hāʻawi koke ʻoe i nā lako i hāʻawi ʻia ma nā wahi he nui, a ʻaʻole ʻoe i ʻike. Eia naʻe, pilikia loa kēia, no ka mea, aia nā mea a pau ma ke kilokilo.
Akā ʻoiai ke hoʻāʻo nei ʻo Shipper e hoʻēmi i ka hopena o ka hoʻopuka ʻana, lawelawe ʻo multi-cluster-scheduler i nā hana maʻamau a ʻoi aku paha ka maikaʻi no nā hana batch.
ʻAʻole loaʻa iā ia kahi mīkini hoʻouna lohi mālie.
Hiki ke loaʻa nā mea hou aʻe e pili ana i ka multi-cluster-scheduler ma .
Inā makemake ʻoe e heluhelu e pili ana i ka multi-cluster-scheduler i ka hana, loaʻa iā Admiralty — nā kahe hana, nā hanana, CI a me CD Kubernetes.
Nā mea hana ʻē aʻe a me nā hoʻonā
He hana paʻakikī ka hoʻohui ʻana a me ka hoʻokele ʻana i nā pūʻulu lehulehu, a ʻaʻohe hopena ākea.
Inā makemake ʻoe e ʻimi hou aku i kēia kumuhana, eia kekahi mau kumuwaiwai:
- He mea paahana ia e hoʻopili ai i nā ʻupena i uhi ʻia o nā pūʻulu Kubernetes like ʻole.
- Hoʻohana ʻia ke kaulahao kūʻai ʻo Target .
- E ho'āʻo e hoʻohana i ka IPV6 a .
- Hiki iā ʻoe ke hoʻohana i kahi mesh lawelawe, no ka laʻana .
- Hāʻawi ʻo Cilium, kahi mea hoʻohui pūnaewele pahu pahu , hiki iā ʻoe ke hoʻohui i kekahi mau pūʻulu
ʻO ia wale nō no kēia lā
Mahalo iā ʻoe no ka heluhelu ʻana a hiki i ka hopena!
Inā ʻike ʻoe pehea e hoʻopili pono ai i nā pūʻulu lehulehu, .
E hoʻohui mākou i kāu ala i nā loulou.
Mahalo nui iā Chris Nesbitt-Smith () a me Vincent de Sme () (ʻenekinia hilinaʻi ma ) no ka heluhelu ʻana i ka ʻatikala a kaʻana like i ka ʻike e pili ana i ka hana ʻana o ka hui.
Source: www.habr.com