みなさんこんにちは。私たちの話を聞いてくれてありがとう。私たちはここで幸せです。では始めましょう。今日は、新しいキャプターをお話しします。リブランディングと新しいリリースタイクのLTS長期のサポートについてお話しします。私たちのサポートをご紹介します。私たちはノムフルエンティープロジェクトです。私は大地郎福田です。そして私はケンタロウハヤシです。私たちはクリアコードのソフトウェアエンジニアルです。クリアコードは日本のソフトウェアデビュロクメントコンパニーです。クリアコードはフリーソフトウェアルのフォアやビジネスを使っています。私たちはとても unique と思います。私たちはフリーソフトウェアデビュロクメントコンパニーです。このグラフは、Github Repositoryのノムフルエンディーでアクティブしています。この2年です。3のノムフルエンティープロジェクトはクリアコードです。私たちはノムフルエンディーの新しいデビュロクメントコンパニーです。3のノムフルエンディープロジェクトは私です。4のノムフルエンディーは天下やしです。今日は、4のノムフルエンディーのashie エンジニアルはここではないです。この家は、今日は本当に幸せです。今日は、こちらのアジェンダーです。1つの動きは、フルエンディーとの方法です。2つの動きは、ノムフルエンディーの歴史です。So how FluentD was developed and how the Fluent package was born.And third topic is what new Fluent package LTS.So we released the new package Fluent package this year.So we're going to explain it in details.Then fourth topic is remarks of updating.So how users can update to Fluent packagefrom TD agent and why users should.And last but not least, we're going to talk about major changesand the future of FluentD.At the end of this talk, we will be happy if everyone here understands about FluentD and new packageand feel encouraged to use them.And more people interested in the FluentD community.So let's start with what is FluentD.I'm sure everyone here heard about FluentD.No, FluentD?Yes.The need to collect data continues to expand.And FluentD is one of known data collectors.So FluentD is an open source data collectorthat unifies data collection and consumption.It is set in official website and it is used by so many companiesand also for over 1300 times in GitHub.So FluentD is a very famous product.So here is an image of data collectionwithout FluentD.So in recent use cases, there are many data sourcesand data outputs.So data sources are like log files and syslogand so onand data comes from many kinds of sourcesand in different formats.And data outputs are like files, databasesand search enginesand so on.And it requires to be a specific format.So without FluentD, it will be very complex like this.But FluentD solves it.FluentD unifies logging layerand organizes data processing.So FluentD can do a whole set of thingsthat are important for data processing.Collecting and filtering and formattingand buffering and routing and sending.So as you can see, data collectionwill be very simple with FluentD.So there are three remarkable pointswhy use FluentD.First, flexibility.So FluentD has plug-in system.And plug-in system unifies many data sourcesand data outputs.And second, scalability.So FluentD can handle large data collectionand also it is well compatiblewith distributed systemslike cloud native computing.And third, open source and community.So anyone can participate in the developmentof FluentD.So first, flexibility.So FluentD has a plug-in system.And there are over 1,000 plug-insfor many data sources and outputs.So you can chooseand combine themand you can achieve what you want.And if you cannot findany plug-in you want,then you can make your own plug-inswith Ruby.Do you love Ruby?No, do you love Ruby?Yes, it is very convenient.And second, scalability.When FluentD work as a folderand aggregators in the log system,you can change the numberof folders and aggregatorsdepends on how much data to transfer.So how to manage it is all up to what you want.And the number of instancescan be dynamically changed.So this scalability is very importantfor FluentD.And third, open source and community.So FluentD is open source.The license is a part license 2.0.So you can refer to the implementationand you can make your own plug-in.Also, if you find any bugs,you can report it.And some bugs are fixedwith the community's help.So you can join the developmentof FluentD.And if it is necessary,the community supportsto maintain your plug-inswith nursery project.So you can get helpto maintain your plug-ins.It's now you know what is FluentDand why you use it.So next,I'm going to tell you the history.FluentD was born on2011 at TreasureData.At TreasureData is United Statesorigin company providing themarketing data platform.And version 1 is releasedon 2017.So it was the base of the current product.And also,FluentD was acceptedto Cloud Native Computing Foundation,CNCF,on 2016.And on 2019,it becamethe graduated maturity level.So it is a great thing.So it was good to have supportfrom CNCF for FluentD.So now,I'm going to explainthe history of the package.So distribution packageof FluentD has also been developedfrom the early daysof the development of FluentD.So we need a packageto use FluentD more easily.So FluentD doesn't workwithout any plug-insin the practical use cases.So users need to installthe plug-ins for specific features.And also,FluentD requiresruby environment.So if there is a package,including the often used plug-insand ruby,it is very convenient for users.So the package of FluentD has been developed.So here is the history of the package.As a package,a TD agent was developedby TreasureData 2011.So originally,the packagewas for TreasureData'sown services.But it became widely usedas a general packageof FluentD.As you can see,along the development of FluentD,a TD agent had been developed too.And recently,the development becamethe community basis.And this year,we releasedFluent packageas a successor of TD agent.So why we changed the nameand what else has changed?So why we changed the name?It's actually,it was a bigchange we had made.The reason of this change isgenerally,the name of the packageis better to describe its main software.And the name,TD agentdoes not associate with FluentD.So it is difficultto imagine that TD agent isactually the package of FluentD.And also,TD agent is so-calledall-in-one package,so not limited for TreasureData's servicesanymore.So the new package name,Fluent package,reflectsthe community-driven development.So it is importantto make the namedescribing,it includes FluentD.So we have discussed with the communityand make the nameFluent package.So it was a big challengeto change the name for developers.So we have worked throughto make sure it does not causemany troubles for users.So finally,we made it.We made it.Thank you,thank you.So,you know,it's not easyto change the name.I'm gonna share details later.Yes.And also,while changing the name,we discussed with the communityabout the new release cycleas well.So I have introducedwhat is FluentDand the history of it.Now you realizethe name have changed.So what else has changed?Not only did we change the name,but also we changed therelease cycle.That is the Fluent Package LTS.So now let me sharewhat is new,the Fluent Package LTS.So LTS means Long Term Support.The newest version,version 5.0.2had been released at this November.So it is recommendedfor stable production use.So while we are involvedin the development of FluentD,there are certain number of requeststo have longer support life cycle.So in the LTS package,we provide scheduled life support life cycle.And in that life cycle,we provide only the security fixesand bug fixes.So it makes it easyto update continuously.So we believe thatthis will meet the needs of users,users who want to useFluentD more stablyover the long term.So I'm going to explain the life cycle in more detail.So there are two channelswith different life cycles.A standard and LTS.A standard is the same life cycleas a TD agent.So it is very simple.About every three months,we update the emulated FluentDto the latest version.So you can always usethe latest functionalitieswith standard channel.But some users may feelit is difficult to update.So because they mayworry aboutencountering new bugsor losing compatibility.So on the other hand,LTS life cycle is for two yearsor so.And we continue to provideonly security fixesand bug fixes.So it means this possibilityofencountering new bugsor losing compatibility.So it is recommendedfor stable production use.And here is the chartto describe the schedule image.So first section describethe scheduleof TD agent version 4 series.And it is already announcedTDAgent version 4 seriesreaches end of lifeby the end of 2023.So this year.And the middle partwith a white backgroundwith a standard channeland the bottom partwith yellow backgroundis the LTS channel.And it is the same packageversion 5.0.But it will changein the future.So as you seein standard channelwe update the minor version.Sorry.I hope.Currentlyboth are providedthe same package version 5.0.But in the futurein standard version5.1when the next minorversion updateversion 1.17Yes.So this is different from LTS channel.Yeah.In the LTS channelwe continues to provideversion 5.0 seriesat least for two years.Yes.And weannounce the major updateto be scheduledin advance.So at the momentthe next major updateversion 6will be released on2025.So thecurrent version5.0 will be supporteduntil then.So let's summarize the pointsof the front package LTS.First secureyou can easily continue updating.And secondless hassle for major updates.So we announcethe support lifecycle in advance.So it makesmeasure updates more user friendly.So user doesn'tneed to be hassles.We hassle miss it.Okay.So Mr. Hayashi will take overthe rest of the presentation.Andthank you for Fukuda.Then I'm going to take overthe rest of the presentation.I'mmary connected to the front packagethe development of the front package.So fit to explain it.Hearing what he had toldnow I'm sure you are readyto use new front package.Now we're going to hearsome points that what need to bebacked for update.This informationwould be useful for you.Many of youmany of you might wonder how difficultto update.In most cases,you just install front package.Just install withinstallation script.It's ready to use it.You can feel how easy it would beto update.But my newmary is stillnecessary because tddcan be customizable path in many ways.We can expectall those migration pathsready-minor migration.Basicallywe take care of keepingcompatibility.So if you upgrade from tddinviron 4,you cansee use all commandall config file and commandnames,usernamegroups.If you have everexplained package developmentdivin packageor alpine package,youknow the difficulty of thesetransitional package.The internal passwordchanged and forpropernet all command,itneeds to make symbol linksfor moving allraw files.It needs to implementmeta hook script.Usingmeta hook scriptmust be carefully implementedand not breakpackage at all.For usabilityallraw rotationneeds to keep all username and group.If you installnuclear reference packagethis compatibility layeris not provided.In general,specific tore-nextand specific tore-window.We already publishedacquired guideas rentedblog article.In general,you install additionalprograms.You need to install programsfor print package.Because the program path oftd is not recognizedby print package.This is by doing the printdiagnostic to gemsupport to detectonly list of binaryrocker install gems.So,it will helpto reinstall gems for print package.And you needto update the file pathsetting if you monitoringprint this log file.For re-next,you need to care new path of printpackage.Aberman file,roborot disk configuration,and system unit file.As for system unit file,if you customize it,need to update printbinary process and someAberman library toolfor windows.If you customize command-writeargument,you need to set them againbecause print package willinstall or reset them.And not that since printpackage version 5, printd service does not startautomatically.There is no merit to startautomatically because you needto shape configuration inadvance.On windows,there is one more thing for you.It is not limited toprint package changeand since it isversion 4.4.2,system directivewas supported.So you need notto customize command-writeargument for log-rotateconfiguration.Just use system directiveinfinity configuration.This should make your systemmeters easier becauseno need to fixlog rotation optionYou can explainremark topic about generalplatform specific ones.If you have any problemproud of the new system,please let us know.Writing communitycan support you.Next topic is major changes.We are going to sharetechnical details of new version.This point was coveredin previous section.But let me explain again.If updated fromtd.agent.parry4,command name was changedto frontd or frontgen.But you can still useall command name.And the name of the systemthe unit file was alsochanged to frontd.You can still usetd.agent as systemd's AES.On windows,frontd service doesn't startautomatically.There are manythings for frontpackage.Remarkable changes are backfixedofintel.infrontdand updated withbindedrelivi version.This is the screenshot ofrenty-resnote.There was a wrongstanding bug whichwrongly stopping tearingintel.Intel is used to correctthe abandoned content of log files.This bug was observedtime to time.It was fixed.Something happened and not happened.This bug was fixedin recent release of frontd1.16.2and 1.16.3withcollaboration of many users.The calls were foundand it was fixedheand the chain number onecontributor of frontd.Dang it!This is the screenshot ofgithub issue about this bug.Feedback from uservery appreciate.This screenshot of suggested fix.We could fix itbecausewith this kind of feedback.With that,I don't know when we could fix it.I don't know.Reveal 2.71Reach and arrive.Frontpackage rendersRevealReveal 3.2processor with performance improvementYou can take advantage of it.Though it is still experimentexperimental in frontdbut I'm sure you knowyou can try wise it.If frontd developercould correctenough positive feedbackWise it will be enabledvery default in someday.We have explained current workand it is importantfor the committee to think about the future.In order to plan futureunderstandingthe current portion isnecessary.So,here are some facts from statisticsof frontd activities.From these statisticswe may see whathad happened and whatwill happen.First,I will introducerecent top file requesttagent frontpackagereport history on AWS Mostrequest comes fromunited states.Followed byNezurans,Japan,Rans,and arrive.In Japan,tagent frontpackageare used more than we expected.Compared between before LTSand after LTS.The access from Taiwan isgraduallyincreased.It is curious factbut I don't know the reason why.If anyone knows the reason,please stay with us later.We foundand deserve all facts.Most of youstill use version 3or version 4.Version 3 are allammented.It's surprisedbecause there is still an access for version 2.It's old, old, old stable.It should not be used.We found that popular packageis td-agent version 4for Amazon renex.Td-agent packageis provided forinter architecture.But maybe with cost meritam architecture was selected.Currentlygithub discussion has low activity.Many questions werepossed butthere's a few number of supporters.So we want to encourageusers to help each other.The feature.How do you feel about the fact ofstatistics?First one,we interest LTSfor enterprise use.So don't keepusing old package.Second one.Currentlyit adds to more community-baseddevelopment.But virtually hostingis heavily rely on treasure data.It's hard to owecurrently package a currentproject about massive traffic.So we contactCNCIF for sponsoring.CNCIFis supportive about it.Third one.There's a difficultyto get help in timeremembering ingithub discussions.Please more and more people join it.When fixing forinter bug,many people give a speed up.So we'd be happyif same thing will happenon github discussion too.Last but not least,release channel of LTShas just started.We want to keepcontinuous releasewith your feedback.Here's a wrap-up.If you're excited to knowfriend package LTSand want to use itor want to joinfriendly community,very appreciate it.Thank you for the program.We'd like to talk afterwardsif you have some feedback.Thank you.Thank you for good presentation.I have a question.Yes.I am currently using friendlybut I know thatI think they arevery similar package.Friendly bit.So when I knowI knew the friendly bit,I had thought thatfriendly will bedeprecated in near futureit will bedeprecated to friendly bitbut it seems thatyou guystry to doaggressivework for friendly.So I would like to knowis there a plan todeprecate friendlyand replace to friendly bitor you guyswill continue todoaggressive work for friendlyThank you for your question.Yes, there isfriendly bit.So it is more lightweight versionof friendly.But it will beaggressive with friendly we think.This is an example ofdata collection systemand friendly bitis more lightweight.So we use itsome forwarderstoin whichwe can't use more resourcesyes, butsome environmentwe can use more resourceswe can use friendlyand the aggregatorswe can use friendly tooand the merit of friendlyis there are many pluginsand we can make pluginswith Ruby.It is easy to make pluginsso we can easilyexpand the functionalitywith what you wantso it will merit of friendlyso we thinkthe friendly and friendly bitwill co-exist in the future.Okay, thank you.I understand that we can use both of friendly and friendly bitfor a long time.Yes, thank you.