Slurm: kab ntsig tig mus ua npauj npaim

Slurm: kab ntsig tig mus ua npauj npaim

  1. Slurm yeej tso cai rau koj nkag mus rau hauv lub ntsiab lus Kubernetes lossis txhim kho koj txoj kev paub.
  2. Cov neeg koom nrog zoo siab. Tsuas muaj qee tus ntawm cov uas tsis tau kawm dab tsi tshiab lossis tsis tau daws lawv cov teeb meem. Lub unconditional moneyback ntawm thawj hnub ("Yog hais tias koj xav tias Slurm tsis haum rau koj, peb yuav refund tag nrho cov nqi ntawm daim pib") tsuas yog siv los ntawm ib tug neeg, justifying tias nws overestimated nws lub zog.
  3. Tom ntej no Slurm yuav muaj nyob rau thaum ntxov Cuaj hlis nyob rau hauv St. Selectel, peb tus neeg txhawb nqa mus tas li, tsis yog tsuas yog huab rau sawv ntsug xwb, tab sis kuj yog nws chav sib tham.
  4. Peb tab tom rov hais dua qhov yooj yim Slurm (Lub Cuaj Hli 9-11) thiab nthuav qhia qhov kev pab cuam tshiab: DevOps Slurm (Cuaj Hlis 4-6).

Slurm yog dab tsi thiab nws tau hloov li cas?

Ib xyoo dhau los, peb tuaj nrog lub tswv yim ntawm kev coj cov chav kawm ntawm Kubernetes. Thaum Lub Yim Hli 18, Slurm-1 tau tshwm sim: nyuaj, nrog kev nthuav qhia txuas ntxiv (thaum qhov kev nthuav qhia tiav rau theem), nrog rau cov teeb meem txhua hnub. Kev sib koom ua ke: cov neeg koom nrog thawj Slurm, zoo li Kev Sib Koom Tes ntawm Lub Nplhaib, tseem sib txuas lus nrog ib leeg.

Slurm: kab ntsig tig mus ua npauj npaim
Qhov no yog qhov Slurm-1 zoo li

Thaum thawj Slurm, lub tswv yim ntawm kev tuav lub MegaSlurm tau yug los. Peb nug tib neeg txog cov ncauj lus uas lawv xav tau, thiab thaum Lub Kaum Hli peb tau muaj chav kawm qib siab "Los ntawm kev thov ntawm cov neeg koom." Nws hloov tawm los ua ib qho kev nthuav, tab sis ib zaug. Txog rau lub Tsib Hlis 19 peb tau npaj cov chav kawm siab heev, nrog rau nws tus kheej cov logic thiab keeb kwm sab hauv.

Nyob rau hauv lub xyoo, Slurm tau hloov lub koom haum:
- Docker thiab Anisble raug tshem tawm ntawm txoj haujlwm tseem ceeb thiab ua cov chav kawm online cais.
- Peb tau teeb tsa kev txhawb nqa uas pab cov tub ntxhais kawm daws teeb meem kev kawm pawg.
- Cov neeg hais lus tam sim no muaj kev txhawb nqa txheej txheem.

Slurm: kab ntsig tig mus ua npauj npaim
Pab neeg uas ua Slurm 4

Cov lus teb los ntawm cov neeg koom

Lwm cov ntaub ntawv tau teeb tsa: 170 tus neeg koom ntawm Slurm hauv paus, 75 ntawm MegaSlurm.

Slurm: kab ntsig tig mus ua npauj npaim

Lus - 4
101 ntawm 170 tus neeg ua tiav daim foos tawm tswv yim.

Puas yog Kubernetes pom tseeb?
41 - Kuv tsis to taub k8s tseem, tab sis kuv pom qhov twg yuav khawb.
36 - Kuv tsis paub k8s ua ntej, tab sis tam sim no kuv tau xam nws tawm.
23 β€” Kuv paub k8s ua ntej, tab sis tam sim no kuv paub zoo dua.
1 - Kuv tsis tau kawm dab tsi tshiab.
0 - Kuv tsis tau nkag siab dab tsi txog k8s.

Koj nyiam qhov kev siv ntawm Slurm li cas?

16 Cov neeg xav tias Slurm yooj yim dhau thiab qeeb, thiab 14 cov neeg xav tias nws nyuaj dhau thiab nrawm. Tsuas yog rau tus so.

Koj puas tau daws qhov teeb meem koj yuav mus rau Slurm nrog?

90 - Yog.
11 - Nr.

MegaSlurm

40 tus neeg tau sau daim foos tawm tswv yim. 2 tus neeg hais tias yooj yim heev thiab qeeb. 1 tus neeg tsis daws qhov teeb meem uas nws mus rau Mega. Tus so yog OK.

Kev tshuaj xyuas ntawm Slurm ntawm https://serveradmin.ru

Hais lus tshuaj xyuas

Slurm: kab ntsig tig mus ua npauj npaim

Yog hais tias nyob rau hauv St. Petersburg Slurm nyob rau hauv Lub ob hlis ntuj muaj feem ntau beginners, ces nyob rau hauv lub Moscow Slurm neeg nyob rau hauv loj tus naj npawb twb sim Kubernetes. Muaj ntau cov lus nug siab heev uas ua rau koj xav.

Yog hais tias nyob rau hauv St. Petersburg lawv nug thaum twg peb yuav luam tawm peb diav rawg ntawm kubespray, ces nyob rau hauv Moscow lawv twb nug yog vim li cas peb thov kom siv peb diav rawg thiab tsis coj tus qub kubespray. Qhov no yog qhov kev xav tseem ceeb ntawm cov neeg laus nruab nrab.

Kev xyaum yog qhov nyuaj, tib neeg ua yuam kev ntau, thiab qhov ntawd yog qhov zoo: koj yuav tsum ua yuam kev thaum kawm, thiab tsis yog hauv kev sib ntaus sib tua.

Peb niaj hnub ntsib kev txwv ntawm kev tau txais daim ntawv pov thawj, txwv kev rub tawm los ntawm Github, thiab lwm yam. Qhov no yog lub neej - peb ib txhij xa tawm txog 200 pawg hauv Selectel huab. Tsis muaj leej twg npaj lawv cov peev txheej thiab kev txwv rau qhov no.

Tshaj tawm ntawm Slurm ntawm Selectel

β†’ Sau npe rau Slurm-5
Nqe: 25 β‚½

Qhov kev pab cuam:

Ntsiab lus #1: Taw qhia rau Kubernetes, cov khoom tseem ceeb
- Taw qhia rau k8s thev naus laus zis. Kev piav qhia, kev siv, kev xav
- Pod, ReplicaSet, Deployment, Service, Ingress, PV, PVC, ConfigMap, Secret

Ntsiab lus No. 2: Cluster design, main components, fault tolerance, k8s network
-Cluster tsim, lub ntsiab Cheebtsam, txhaum kam rau ua
- k8s network

Ntsiab lus #3: Kubespray, kho thiab teeb tsa Kubernetes pawg
- Kubespray, teeb tsa thiab kho ntawm Kubernetes pawg

Ntsiab lus #4: Advanced Kubernetes Abstractions
- DaemonSet, StatefulSet, RBAC, Txoj Haujlwm, CronJob, Pod Scheduling, InitContainer

Ntsiab lus #5: Publishing services and applications
- Txheej txheem cej luam ntawm cov kev pabcuam tshaj tawm: NodePort vs LoadBalancer vs Ingress
- Ingress maub los (Nginx): ntsuas cov tsheb thauj mus los
- Tus thawj tswj hwm: tau txais SSL / TLS daim ntawv pov thawj

Ntsiab lus #6: Taw qhia rau Helm

Ntsiab lus #7: Txhim kho daim ntawv pov thawj-tus thawj coj

Ntsiab lus #8: Ceph: "ua li kuv ua" installation

Ntsiab lus #9: Kev txiav txim thiab saib xyuas
- Kev saib xyuas pawg, Prometheus
- Cluster logging, Fluentd/Elastic/Kibana

Ntsiab lus #10: Pawg hloov tshiab

Ntsiab Lus No 11: Kev ua haujlwm zoo, kev siv dockerization thiab tso rau hauv ib pawg

Cov chav kawm ntawm Docker thiab Ansible ntawm stepik.org suav nrog tus nqi.

β†’ Sau npe rau Slurm DevOps
Nqe: 45 β‚½

Qhov kev pab cuam:

Ntsiab lus #1: Taw qhia rau Git
- Basic commands git init, cog lus, ntxiv, diff, cav, xwm txheej, rub, thawb
- Teeb tsa ib puag ncig hauv zos: cov lus pom zoo
- Git ntws, ceg thiab cov cim npe, sib koom ua ke cov tswv yim
- Ua haujlwm nrog ntau qhov chaw taws teeb repo

Ntsiab lus # 2: Pab neeg ua haujlwm nrog Git
- GitHub ntws
-Fork, tshem tawm, rub thov
- Kev tsis sib haum xeeb, kev tshaj tawm, ib zaug ntxiv txog Gitflow thiab lwm yam dej ntws hauv kev sib raug zoo nrog pab pawg

Ntsiab lus #3: CI/CD qhia txog automation
- Automation hauv git (bots, taw qhia rau CI, hooks)
- Cov cuab yeej (bash, ua, gradle)
- Lub Hoobkas sib dhos kab thiab lawv daim ntawv thov hauv IT

Ntsiab lus #4: CI/CD: Ua haujlwm nrog Gitlab
- Tsim, sim, xa tawm
- Cov theem, hloov pauv, tswj kev ua tiav (tsuas yog, thaum, suav nrog)

Lub ntsiab lus # 5: Ua haujlwm nrog daim ntawv thov los ntawm kev txhim kho qhov pom
- Peb sau microservice hauv Python (nrog rau kev xeem)
- Siv docker-compose hauv kev txhim kho

Ntsiab lus #6: Infrastructure li Code
- IaC: mus kom ze rau infrastructure li code
- IaC siv Terraform ua piv txwv
- IaC siv Ansible ua piv txwv
-Ideempotency, kev tshaj tawm
- Xyaum tsim Ansible playbooks
- Configuration cia, kev sib koom tes, daim ntawv thov automation

Ntsiab lus #7: Infrastructure testing
- Kev sim thiab txuas txuas ntxiv nrog Molecule thiab Gitlab CI

Ntsiab lus No. 8: Automation ntawm raising servers
- Sau cov duab
- PXE thiab DHCP

Ntsiab lus #9: Infrastructure Automation
- Ib qho piv txwv ntawm kev pabcuam kev tsim kho vaj tse rau kev tso cai ntawm servers
- ChatOps (kev koom ua ke ntawm instant messengers nrog pipelines)

Ntsiab lus #10: Kev ruaj ntseg Automation
- Kos npe CI/CD cov khoom qub
- Kev ntsuas qhov tsis zoo

Ntsiab lus #11: Saib xyuas
- Lub ntsiab lus ntawm SLA, SLO, Kev Siv Nyiaj Txiag thiab lwm yam lus txaus ntshai los ntawm lub ntiaj teb ntawm SRE
- SRE: SLI thiab SLO saib xyuas kev xyaum
- SRE: Kev xyaum siv nyiaj txiag yuam kev
- SRE: cuam tshuam thiab kev tswj xyuas kev ua haujlwm (apigateway, kev pabcuam mesh, Circuit Court breakers)
- Saib xyuas cov kav dej thiab cov ntsuas kev txhim kho

Tau qhov twg los: www.hab.com

Ntxiv ib saib