Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
メルカリのシステム・サービス監視について/Monitoring Mercari service...
Search
kazeburo
November 29, 2017
Technology
5
7.1k
メルカリのシステム・サービス監視について/Monitoring Mercari service and servers
メルカリのシステム・サービス監視について
Monitoring seminar in Mercari
kazeburo
November 29, 2017
Tweet
Share
More Decks by kazeburo
See All by kazeburo
クラウド開発の舞台裏とSRE文化の醸成 / SRE NEXT 2025 Lunch Session
kazeburo
1
450
さくらのクラウド 開発の挑戦とその舞台裏
kazeburo
1
950
[SRE kaigi 2025] ガバメントクラウドに向けた開発と変化するSRE組織のあり方 / Development for Government Cloud and the Evolving Role of SRE Teams
kazeburo
4
3.2k
[さくらのTech Day] ガバメントクラウド開発と変化と成長する組織 / sakura techday, Develop govcloud and the team
kazeburo
0
6.8k
ガバメントクラウド開発と変化と成長する組織 / Organizational change and growth in developing a government cloud
kazeburo
4
3k
DNS水責め攻撃と監視 / DNS water torture attack Monitoring and SLO
kazeburo
5
4.3k
DBやめてみた / DNS water torture attack and countermeasures
kazeburo
13
14k
IaaSにおけるPlatform Engineeringとこれから / Platform engineering in IaaS
kazeburo
2
1.4k
高信頼IaaSを実現するDevOps / DevOps for Highly Reliable IaaS
kazeburo
1
710
Other Decks in Technology
See All in Technology
United Airlines Customer Service– Call 1-833-341-3142 Now!
airhelp
0
180
NewSQLや分散データベースを支えるRaftの仕組み - 仕組みを理解して知る得意不得意
hacomono
PRO
3
230
話題の MCP と巡る OCI RAG ソリューションの旅 - Select AI with RAG と Generative AI Agents ディープダイブ
oracle4engineer
PRO
5
110
cdk initで生成されるあのファイル達は何なのか/cdk-init-generated-files
tomoki10
1
540
Zero Data Loss Autonomous Recovery Service サービス概要
oracle4engineer
PRO
2
7.8k
[ JAWS-UG千葉支部 x 彩の国埼玉支部 ]ムダ遣い卒業!FinOpsで始めるAWSコスト最適化の第一歩
sh_fk2
2
150
Copilot coding agentにベットしたいCTOが開発組織で取り組んだこと / GitHub Copilot coding agent in Team
tnir
0
150
Rethinking Incident Response: Context-Aware AI in Practice
rrreeeyyy
1
390
How to Quickly Call American Airlines®️ U.S. Customer Care : Full Guide
flyaahelpguide
0
240
AWS CDK 入門ガイド これだけは知っておきたいヒント集
anank
5
600
ポストコロナ時代の SaaS におけるコスト削減の意義
izzii
1
260
Enhancing SaaS Product Reliability and Release Velocity through Optimized Testing Approach
ropqa
1
250
Featured
See All Featured
CoffeeScript is Beautiful & I Never Want to Write Plain JavaScript Again
sstephenson
161
15k
The World Runs on Bad Software
bkeepers
PRO
69
11k
For a Future-Friendly Web
brad_frost
179
9.8k
CSS Pre-Processors: Stylus, Less & Sass
bermonpainter
357
30k
Fashionably flexible responsive web design (full day workshop)
malarkey
407
66k
The Language of Interfaces
destraynor
158
25k
How GitHub (no longer) Works
holman
314
140k
Bash Introduction
62gerente
613
210k
Measuring & Analyzing Core Web Vitals
bluesmoon
7
510
YesSQL, Process and Tooling at Scale
rocio
173
14k
ReactJS: Keep Simple. Everything can be a component!
pedronauck
667
120k
RailsConf 2023
tenderlove
30
1.1k
Transcript
ϝϧΧϦͷγεςϜɾαʔϏε ࢹʹ͍ͭͯ Monitoring Seminar in Mercari 2017/Good/Meat @kazeburo
Me • Masahiro Nagano / խ • id:kazeburo • Mercari,
Inc Principal Engineer Site Reliability Engineering (SRE) Team
Agenda • Mercariͷ͜Ε·Ͱͱࢹπʔϧ • MackerelͰͷαʔόࢹ
~ 2014/9 JP ੴङ ΠϯϑϥνʔϜ!5PLZP
~2014/9 • ͘͞ΒΠϯλʔωοτੴङDCͷઐ༻αʔόͱΫϥυΛར༻ • ઐ༻αʔόʹͯZabbixαʔόΛߏங • ʮτϦΨʔʯΛ׆༻ͯ͠͞·͟·ͳࢹΛߦ͏ • ͍݅ࣜͩ͠Ͱෳࡶͳࢹ͕࣮ݱ •
ݱࡏͷࢹ߲ͷϕʔε͕Ͱ͖Δ
2014/9~ US JP ੴङ ΠϯϑϥνʔϜ!5PLZP
2014/9~ • USαʔϏε͕ AWS Oregon region ʹͯ։࢝ • ઐ༻αʔόͱΫϥυͷϚϧνˍϋΠϒϦουߏ •
USʹZabbix ServerΛߏஙͯ͠ɺ౦ژ͔Βࢹ
ଟRegion Zabbixͷ՝ • Zabbixͷઃఆ͕ͣΕ͍ͯ͘ • ӡ༻͍ͯ͠ΔZabbixͷόʔδϣϯ͕ҟͳΔ • ઐ༻αʔόͱAWSͰएׯҧ͏ࢹ߲ • JPͰ࡞ΓࠐΜͩࢹ͕USͰ࠶ݱͰ͖͍ͯͳ͍
• USͰ͚ͩى͖Δࢹ࿙ΕʹΑΔࣄނ • Zabbix ProxyΛར༻͠ɺ1ݸͷZabbix ServerूͳͲͷҊ
Zabbixͷ՝ • Zabbix ࣗମͷӡ༻ • όʔδϣϯΞοϓͷෛ୲ • MySQL ͷෛՙ͕େ͖͘ɺࢹԆͳͲൃੜ •
ෳࡶͳτϦΨʔͷཧ • ϚεΫϦοΫओମͷઃఆ • όʔδϣϯཧͳͲΛߦ͍͍ͨ • ࢹͷ௨Λվળ͍ͨ͠
2016/1~ US JP ੴङ 43&!5PLZP
mackerel ಋೖ • Service Metrics͔Βಋೖ • ؆୯ʹάϥϑ͕ඳ͚ɺࢹᮢͷઃఆ͕Ͱ͖Δ • fluentdɺNorikraͱͷΈ߹Θͤ •
ZabbixͷτϦΨʔͷҠ২ • τϦΨʔΛPluginͱ࣮ͯ͠ • Plugin GitͰཧ͠ɺAnsibleͰ
mackerel ಋೖ • ࢹπʔϧɺ࣌ܥྻDBͷӡ༻ͷΦϑϩʔυ • Կ͠ͳͯ͘ຖिόʔδϣϯΞοϓ • JP/USͰͷࢹ߲Λ߹ΘͤΔ • ҟͳΔͱ͜Ζ
Ansible templateͳͲͰٵऩ
2017/3~ US JP ੴङ UK 43&!5PLZP
2017/3~ • UK ͰͷαʔϏε։࢝ • UK Λ։࢝͢Δʹ͋ͨͬͯɺ͞Βʹ͏ҰͭͷΫϥυΛ࠾༻ • ࢹ͕ΫϥυԽ͞Ε͓ͯΓɺ৽ͨͳࢹαʔόͷՃඞཁͳ͠ •
JP/US ͷࢹ߲͕ͦͷ··ద༻Ͱ͖ɺΠϯϑϥετϥΫνϟͷߏஙྃͱͱ ʹࢹͷઃఆ͕ྃ
ݱࡏ US JP ੴङ UK 43&&OHJOFFST!+1646, Stackdriver Prometheus
ݱࡏ • ϚΠΫϩαʔϏεԽ • GKE ্ͷίϯςφɾαʔϏεͷࢹͷͨΊʹ Stack DriverɺPrometheusɺ DataDog ͷ׆༻
• αʔόαΠυΤϯδχΞࢹπʔϧΛར༻
ͦͷଞͷࢹ New Relic Kurado άϥϑը૾ΛҰؾʹݟΕΔͷͰศར جຊతͳϝτϦΫεͪ͜ΒͰݟΔ PHPͷ෦ͷτϨʔε ΞϓϦέʔγϣϯͷνϡʔχϯάͷࢀߟ
MackerelͰͷαʔόࢹ
https://speakerdeck.com/kazeburo/mackerel-day
ࢹʹ·ͭΘΔࣈ • ࢹϧʔϧ: 278 • Hostຖͷࢹϧʔϧ • MySQL: 34 •
Application: 39 • Search: 37 • Custom Plugin: 50+ (check + metrics + utils)
MySQLͷࢹ߲(1/4) • Connectivity • FileSystem % >85% >88% • Swap
% >50% >70% • ssh-alive • sshdͷϓϩηεࢹ • global-ip-and-iptables • global ipͷ༗ແͱiptablesͷঢ়ଶ • unbound-resolv • localͷunboudͰ໊લղܾ͕Ͱ͖Δ͔ • unbound-process • unboundͷϓϩηεࢹ • crond-process • crondͷϓϩηεࢹ • uptime • ࠶ىಈࢹ
MySQLͷࢹ߲(2/4) • inode-usage • inode༻ >80% >90% • uname-change •
unameίϚϯυͷ݁Ռͷdiffࢹ • passwd-change • passwdϑΝΠϧͷdiffࢹ • hostname-changed • hostnameίϚϯυͷ݁Ռͷdiffࢹ • custom.ntpq.synced.remote <0.1 <0.1 • custom.ntpq.offset.seconds >300 >300 (msec) • ntpͷಉظαʔόͱ࣌ࠁͷζϨ • custom.linux-lite.memory.avail <50MB <20MB • ۭ͖ϝϞϦ • custom.linux-lite.cpu-usage.cpu-steal >20% >20% • custom.linux-lite.cpu-usage.cpu-iowait >30% >50% • custom.linux-lite-cpu-usage.cpu-system >8% >8% • ͦΕͧΕͷCPU༻(100%্͕ݶ)
MySQLͷࢹ߲(3/4) • cutom.linux-lite.loadavg.per-cpu >3 >3 • ίΞͰׂͬͨϩʔυΞϕϨʔδ • postfix-smtp-alive •
SMTPϙʔτͷ֬ೝ • postfix-master-process • postfix masterϓϩηεࢹ • custom.postfix.mailq.queue >100 >5k • postfix mail Ωϡʔཹ • custom.linux-lite.process.all >2k >2k • custom.linux-lite.process.running >60 >100 • શϓϩηεͱ࣮ߦதͷϓϩηε • mysql-uptime • mysqlͷuptime • custom.mysql-lite.replication-threads.io <0.2 <0.2 • custom.mysql-lite.replication-threads.sql <0.2 <0.2 • ϨϓϦέʔγϣϯͷ֤threadͷঢ়ଶ
MySQLͷࢹ߲(4/4) • custom.mysql-lite.replication-behind- master.second >5 >5 • mysqlͷϨϓϦέʔγϣϯԆ • custom.mysql-lite.connections.utilization
>90 >90 • max_connectionsʹର͢ΔίωΫγϣϯ • custom.mysql-lite.threads.running >1k >2k • mysql্Ͱ࣮ߦதͷεϨου • mysql-slave-sql-error • replicationΤϥʔͷࢹ • machine-exceptions • αʔόͷϝϞϦΤϥʔࢹ • raid-disks • αʔόͷRAID/Diskঢ়ଶͷࢹ
ࢹͷҭ͔ͯͨ • ʮࢹܧଓతͳςετʯby kazuho • ʮςετϑΝʔετʯͰγεςϜΛߏங͠ͳ͕ΒࢹΛ࡞Δ • ϓϩηεࢹϙʔτࢹ • ʮϙετϞʔςϜʯͷରԠࡦͱͯ͠ࢹΛҭͯΔ
• ᮢͷௐɺL7Ϩϕϧͷࢹ
end