Hacker News new | past | comments | ask | show | jobs | submit login
Atlassian products have been down for 4 days (atlassian.com)
384 points by Corrado on April 10, 2022 | hide | past | favorite | 244 comments



I never hear anyone shilling for them, so now is my chance. Anyone who’s looking for a great JIRA alternative, I love Shortcut (formerly Clubhouse). It’s the best task management software I’ve used aside from the “no task management software” option. Very snappy, updates in real time, notifications make sense, and no workflow-manipulation shenanigans.


Using Clubhouse was a huge pain.

Feeling that even they don’t understand what all their epics, stories and milestones do haunted me all the time. Page load times up to 10 seconds and totally unusable on a big screen (just insanely slow after window size bigger then on avarage laptops).

Convincing my company to switch to Linear was better than any holiday.


The concept confusion is the one annoying thing about it. Too many nouns and no rigorous definitions of them. I’ve never encountered page load times, maybe we just have fewer tickets at my job.


I liked (but was conflicted) Clubhouse enough to end up on a sales call. The person I talked to said much the same, “story” is more or less arbitrary and iirc, there’s no real concept of “bug”. We went with Jira on the end. It’s fine, and we’re not that special.

Plus no way to open up issue tracking to the public.


I’m more speaking to people using Jira who hate it but are encountering the “there’s no good enough alternative” argument. I’m glad the Linear crowd is speaking up; I had never heard of it, but whatever alternatives are available, I hope people start using them more, because Jira is just some of the most frustrating software we have to use these days, and there’s no good reason task management should be this bad.


FWIW, Restya Core works good for us. Disclosure: We were privileged to be on their private beta.


+1 , Linear has been fantastic for us


I would like to advocate for Linear.app as an outstanding, fast, polished alternative to Jira for teams looking for an alternative.


gitlab issues seem fine to me. some of the stuff jira does for managers isn't there i guess, but if gitlab solves the engineers problem, having to find a separate tool that can create artifacts for a manager seems like a better idea anyway, one tool per audience.


I think this strength of Jira is its biggest weakness. Trying to do everything for everyone and failing either miserably at everything, or just managing some.


I love linear. I'm starting a culinary mushroom farm (small scale) and need to track innoculation/colonisation times and such so I can forecast.

Linear was my go-to and I signed up today. I used it at getweave.com and loved using it there


If they implemented SSO with Office365/AAD it would be a stellar alternative. The GSuite dependency for SSO is damning


They do seem to offer generic SAML SSO, so I'm guessing you could hook it into M365 manually if necessary. We use JumpCloud and I'm interested in seeing how Linear might integrate using SAML.


To elaborate, Linear is probably the best business app I have used for software management ever. For engineers it is fast and fluid and simple with outstanding keyboard shortcuts. For managers it is strongly opinionated with an agile methodology that requires no setup and just works. For teams, Linear makes it fun to stay in sync and track progress towards goals.

One of the best features of Linear is that it inspires my team to build better software, not just tick feature boxes.


Linear is awesome. It gets out of your way and let’s you focus on creating tickets really quickly. The integration with GitHub means engineers rarely need to update the board since tickets sync with your pull request. The keyboard shortcuts are amazing and so is the performance. I’d highly recommend it.


https://prismic.shortcut.com/pricing

Their pricing and feature breakdown across plans is a joke. Milestones, reporting and roadmap are all top tier features? They're base features on a product like this.


Not once have I wanted reporting or roadmap functionalities as an engineer. Those are management functions, and task management software should serve the people performing the tasks (engineers) first and foremost. Milestones, that’s a bit debatable but still doesn’t seem like core functionality when you have labels and epics.


It’s laughable that you think this is an engineer-first product.

Management products are used by engineers LAST. You project management stuff beyond your own Trello/TODO.txt once someone from management tells you to.


That’s the exact misconception that leads to the development of a tool like JIRA. Despite its name, JIRA is a data entry tool first and foremost, and analysis is run a minority of the time. How many engineers vs managers are there at an organization? Many more engineers. So it makes sense that data entry is the heaviest use case.


What the heck really?

I had the complete exact opposite experience. I actually do wish we are still on jira.

it's slow. It's filters make no sense. It's entire flow make no sense.


Filters make no sense? You craft the filters, so if that makes no sense it’s because you didn’t craft the right thing.

What flow? It’s a task management software, not a traffic system. I don’t want it manipulating the flow of anything. It provides swimlanes and that’s more than enough for me. I can move any task anywhere at any time which is exactly what I want out of a software — for it to get out of the way.


> You craft the filters, so if that makes no sense it’s because you didn’t craft the right thing.

That's where you've lost nearly all the people.

Who in their right mind crafts filters. What does this even mean?


"Filter artisans" I suppose.

It is impossible to tell how many times I tried to filter out something in Jira only to have everything since before the times or absolutely nothing and needed to scratch that and go back to the defaults. Or hunt down the email with an exact link to the issue.


Anyone who uses a search engine, tries to find an email, or generally deals with large datasets.


No one "crafts" filters for the purposes of searching or finding an email.

You use simple built in filters like "site:x" or "in: folder". There's no "crafting" or involved.


You type in the word “site:”. That is crafting.


That's not "crafting". And, additionally, the vast majority of people will never use (or even know) about this "crafting".


Craft: “to make or produce with care, skill, or ingenuity.” It means to make something carefully, the same thing done when crafting a search query. That’s the way I meant it anyway. There is nothing particularly difficult about searching in Shortcut. You combine a set of available filters to make a composed filter the same way you search anything. The difference is that in JIRA, there are about 5 different ways to search and the operators are not always obvious.


What it looks like is you keep defending a shitty UX/UI by trying to justify it with inapplicable examples.

> You combine a set of available filters to make a composed filter the same way you search anything

This isn't even remotely similar to "anyone who uses a search engine, tries to find an email". Which brings us back to the original complaint, quote: "It's filters make no sense."

If you "have to combine/craft filters to create a composite filter that you then use to search", you've failed. In exactly the same ways that Jira fails at this task.


I don’t care if you don’t use Shortcut. Use Linear instead, for example, since apparently a lot of people like it in this thread. I had never heard of it before. Main thing is, don’t pretend like JIRA is the only reasonable option because it just isn’t (not saying you were necessarily, but that is what I have a problem with people doing.)


> I don’t care if you don’t use Shortcut.

I never said whether I was or wasn't using Shortcut.

> Use Linear instead, for example, since apparently

I never said whether or not I was looking for an alternative

> don’t pretend like JIRA is the only reasonable option

I never said anything about JIRA except in the following context: "If you <make things complicated>, you've failed. In exactly the same ways that Jira fails at this task." And you are the one who actually mentioned JIRA first.

So how about: you stop pretending I said or implied something I never did?


You accused me of defending something. What actually happened is I asked a series of clarification questions ("Filters make no sense?", "What flow?"), neither of which was answered by the way. Again, I have no dog in whether or not you use Shortcut. So I am not defending anything.

>I never said anything about JIRA except in the following context

That's exactly why I ended my last comment the way I did ("not saying you were necessarily").

> I never said whether or not I was looking for an alternative

Why are you commenting on this thread? The whole point was to suggest an alternative. You never even suggested another way filters could work. As far as I can tell, the mathematical definition of a filter is that it's composed of sub-filters. So please either admit you're a negativity-spouting troll or contribute something positive.


> The whole point was to suggest an alternative.

I responded to this statement: "You craft the filters, so if that makes no sense it’s because you didn’t craft the right thing".

No. No one in their right mind wants to "craft filters". If you require people to "craft filters" or "combine a set of available filters to make a composed filter", you've failed. It doesn't mater if it's Shortcut, or Linera, or JIRA, or any of the dozens of JIRA replacements/wannabes.

That's the thing I'm responding to. An all your examples and allusions only serve to show that you don't understand this simple fact.

For people to want to "combine/craft filters" these filters have to be immediate, smart, autocompleting, and properly predictive. As an example, Youtrack[1] shows how to do that more-or-less properly (though it has other issues). Shortcut can't even complete available states when you type `state:` into the search box.

> As far as I can tell, the mathematical definition of a filter is that it's composed of sub-filters.

Literally no one using project management software cares about mathematical definition of filter. Not a single soul.

[1] https://youtrack.jetbrains.com/issues


> No. No one in their right mind wants to "craft filters".

You should strongly question this assumption because I am a living counterexample. Before I discovered this tool, I wanted to be able to write the filters so badly I even started to develop a tool that allowed this (called Infinitix) that provided only one basic thing: ability to apply filters on a single list of tickets, and then save or share that filter to other members of your org. Regardless of whether you think "you've failed" if you develop this, I'm sure photographers thought the same thing about Photoshop when it was created. (No, this product never fully materialized, luckily I didn't have to build it because I found out about the existence of Shortcut.) It would have been the simplest task management system possible, in a way.

I don't care if the autocompletions are there, what I care about is having control instead of my system administrator or manager having control, and being able to run whatever query at whatever time I want to. In other words, I want to control the tool rather than it control me, and I also want a tool (instead of TODO.txt, the anti-tool approach to task management).


> You should strongly question this assumption because I am a living counterexample.

One counterexample in an ocean of examples doesn't amount to much.

> Regardless of whether you think "you've failed" if you develop this, I'm sure photographers thought the same thing about Photoshop

If only you could have the will to abstain from inapplicable analogies

> I don't care if the autocompletions are there, what I care about is having control

Yeah that's a very shitty control you have. If you really need "mathematical definition of filters", make sure they are, you know, useful. See short video: https://photos.google.com/share/AF1QipOFZ9y-R-tu0XCMfOhqdRW_... (switch to 1080p if needed)


`state:unscheduled` works just fine, I tested it before writing this comment. There's no limit to the variety of analogies you can dismiss if you simply allow yourself to declare them inapplicable. As for control, it sounds good to me. Honestly, not being able to run reporting is a feature from a developer's perspective.


> `state:unscheduled` works just fine, I tested it before writing this comment.

You clearly didn't understand what I showed you. Watch the video again.

> There's no limit to the variety of analogies you can dismiss if you simply allow yourself to declare them inapplicable

Of course. Analogies never work they way you think they work. At best analogies are dubious claims that need to be proven. You equated Shortcuts to Photoshop, and that analogy is not even tenuous. It's false.

> As for control, it sounds good to me. Honestly, not being able to run reporting is a feature

WHo said anything about reporting? Oh wait, no one, except you.


No one said anything about autocompletion either, yet here we are. I'm a bit done here since you are one of these people who refuse to use analogies to inform their thinking. Suffice it to say, many billions of people over the years have found analogies to be a useful endeavor, and nothing about this particular situation is any different. If you really need it spelled out to you, Douglas Hofstadter has a good talk about analogy as a basis of cognition on YouTube.


> No one said anything about autocompletion either, yet here we are.

We were talking about filters, and this is directly related to the filter functionality that you're so intent on defending.

> I'm a bit done here since you are one of these people who refuse to use analogies to inform their thinking.

I don't refuse to use analogies. However, I refuse to use false analogies, and I also refuse to use only analogies to inform my thinking. I prefer facts over analogies.

> many billions of people over the years have found analogies to be a useful endeavor

And once again you're veering into some completely unrelated tangent that no one is talking about.

> Douglas Hofstadter has a good talk about analogy

If only you had the will and comprehension to keep with the topic at hand without resorting to inventing arguments for other people, overusing false and inapplicable analogies etc.


No way is Shortcut snappy, at least for EU users. They used it for my last gig and I mostly avoided it as much as possible because the constant 1-5 second page transitions drove me crazy.


+1, I used it while it was still called Clubhouse and it was very pleasant. Its performance is the biggest thing I miss after changing jobs and having to go back to Jira.


Based on my experience from 2-3 years ago, I would say to avoid it. It would be a decent pick for a really small team without a lot of product management expectation, which would probably be served just as well by something like trello. Example: We wanted to create tasks from Intercom conversations and the only way was a custom implementation. Using their api was not a pleasant experience. Also there were annoying UI bugs that made the product feel unpolished. I ended up using tampermonkey to fix and improve the UI. We ended up switching to jira and it’s 100% an upgrade, we were happy with the change. People shit on it when it goes down, same as github, but then it’s all fine and dandy the other 99% of the time.


> Shortcut (formerly Clubhouse)

I've been semi-tracking them because they're a local NYC Clojure shop and had wondered how they were handling the social network name overlap. Glad to hear they're still around.


What's with people posting "alternatives" to Jira that aren't self hosted? The problem with jira is that it costs $27k/year to self host.


How about YouTrack? It's pretty competent, comes in self hosted and SaaS editions and is very keyboardable.


Leaning towards that or Mantis


Having used self-hosted JIRA before I can tell you with confidence that’s far from the only problem with it.


I just use todoist. Very simple, you can create a bit of structure if you want it, or just keep it very simple as a to-do list & color code priority.

The primary downside is that viewing history (closed tasks) is complete shit. They have an API, and that's the only way of getting history out in any sort of convenient format to review. A very strange oversight.

I'm also not sure how well it would scale to large teams.


Shortcut is great, but their UI is slow and outdated.


How? It seems responsive to me. Most actions result in changing the UI a bit but showing data that’s already preloaded. So no delay waiting for a server roundtrip most of the time.


We do use extensively in a team of 50+ engineers.

Loading an epic may last 10-15 seconds, triggering the buttons to open the pop-ups menu is sluggish, etc.

I think the main problem is that the browser has to render so many elements, most of them are not even related to the page I am opening. For example, opening a story, it loads behind hundreds of cards before opening the "modal" dialog.

These modal dialogs are the most frustrating experience in SC – they close with the ESC key while you are writing, and it already happened the story I was writing was not saved, despite the obnoxious toast alerts saying the contrary. Modals also often breaks "modals menu" in Safari placing them in random places of the UI. They don't use the whole viewport so there's lot of distracting noise in the whole UI.

Some time ago there was some weird cartoonish animations popping out during my work. What a nice idea to consume even more browser resources and get into the middle of your workflow :(


> Update - Restoration work to restore sites is underway and will continue into the weekend. We are taking a controlled and hands-on approach as we gather feedback from customers to ensure the integrity of these site restorations.

I’ve seen failing Kickstarter updates with more information than this.

Anyway the title isn’t wrong per se but it might be more informative to say a subset of tenants using Atlassian’s SaaS products are down.

If you use any of their products on prem … they are not down. And if you’re not one of the affected tenants then your Atlassian products are fine. At work we have Confluence hosted by Atlassian and it’s up.

According to The Register:

> We were also told that the incident affects a relatively small number of Atlassian customers: about 400. That's only 0.18 per cent of the company's 226,000 customers, which isn't much consolation to the several hundred who still can't access their data.


Yeah, on prem is not affected. But they're discontinuing on prem, making customers even more dependent on their hosted services. Bravo!

(My employer will switch away from Atlassian products once they don't offer on prem anymore.)


We just started moving to JIRA after a few years of Gitlab, where gitlab issues feels like nobody has worked on it for a long time.

And Jira has been really great. The sprint boards, concept of mapping statuses to in progress etc have all been great.

My one gripe (and it’s not a small one) is performance. It still feels quite sluggish. Is Atlassian ever going to address this?


No, consider that they have an entire team dedicated to their WYSIWYG comment text field and it weighs in at over 20mb(!!!) of JavaScript for a text box.

Atlassian absolutely doesn't care about performance, it should've been obvious when they rolled out "new Jira" a couple of years ago and it was just as slow.

Their priority is adding new esoteric settings that their top customers demand, hence why their settings tab is a labyrinth of spaghetti complexity.


Just as slow?

I’ve been using Jira for almost a decade now. Pretty sure it gets slower and slower.


It got MUCH slower about... 7 years? back when it went all javascript-y; it seems to have mostly slowly improved since then, though it's still pretty slow.


I have no expectation of it; They do everything slowly, regardless of how simple the task, and it’s been that way for ages. And they do ridiculous amounts of configuration.

Im fairly sure at this point it’s just mountains of intermixed logic and death by a thousand cuts; I doubt they could significantly improve performance to approach something reasonable even if they wanted to.


> My one gripe (and it’s not a small one) is performance. It still feels quite sluggish. Is Atlassian ever going to address this?

oh my sweet summer child


haha laughed hard at this one.


What features are you missing from Gitlab?

I've used several bug trackers including JIRA. Gitlab has been my best experience.


I want to look at sprint progress by tickets marked as “complete” in a certain status.

But the reports only do it by open and closed. We only merge/close when we’re releasing the ticket.


If you're on cloud there's really nothing you can do; on server there are/were some performance tricks you could do by analyzing the SQL queries and designing the fields to be as slim as possible.


Your brain must adjust to a new reality :)


Depends what type of project you have. If you're on the next-gen, team-managed project, that was written from the ground up so is supposed to be faster.


cloud or onprem? I found cloud to never match what we could get in-house,


cloud. I last used JIRA in 2007 with an on prem install.


Isn’t this the sort of thing that would normally have made it to the first page here a while ago?

Just asking, I’ve never used any of their products


The main thing with Atlassian is they royally screwed over all their customers by discontinuing their most popular product and forcing everyone to cloud using really, really unsavory tactics. The only product remaining you can use without cloud had its price increased by an order of magnitude. And, they offered a discount to on-prem customers who signed up for it, and after the fact announced a price increase at least equal to the discount they were offering.

They have decided to treat their customers worse than Oracle does, with nothing but threats that you can make the pain stop if you just go with cloud. They’re an absolutely despicable company.

So after enduring all this, there’s a huge amount of schadenfreude that customers are experiencing watching them be down for so long. They absolutely deserve it.


Never fear: these funds are being directed to critically useful tasks of an earth-shattering priority like building a wooden skyscraper.


I thought this was a joke or something but after googling it you aren't joking....

Wow...

https://architectureau.com/articles/worlds-tallest-hybrid-ti...

Hell of a bet on building more physical space with WFH and other factors making office space cheaper. Plus investing in a new building material concept. This seems like a passion project for someone high up in their food chain.


Holy... completely with a "habitat" bullshit.

> expected to cost $546 million and be completed in 2026.

> Construction is expected to create 2500 additional jobs and generate $1 billion dollars in revenue annually.


BTW, $546000 / 5y / 4000 persons = $27300 annually. But that wouldn't magically generate $1B/y out of thin air.


Ok but wooden skyscrapers are extremely cool. As a non-customer of Atlassian I'm glad my city will be getting a new landmark made of wood.


Interesting. My knee-jerk reaction was "what a decadence". Then again, it might benefit somehow, just wondering how...

A quote from the project page: "This project is not just about being green, it’s a whole systems approach to design, construction and habitation. It will demonstrate that change is possible and has many advantages for the planet, our cities and business."

Sounds vague.


What was the discontinued most popular product? Hipchat? That was pretty terrible by modern standards.


Orev may be referring to Atlassian sunsetting their Server products. You now have either Cloud or Datacenter ($$$) as your only options.


On-premise for Jira, confluence, etc.

You can't self host now, just use their cloud offerings.


You can self-host, that’s the data center versions.

Upside, it’s the same code just difference license file, and it enables clustering (plus lots of other artificially limited features)

Downside, it’s a subscription license and it’s insanely expensive.


I would have thought so, but I think it's because people have low expectations of Atlassian that it hasn't gained traction.

In ranking companies by "Ugh, your product sucks" many people would put Atlassian pretty high. So if their product sucks, why would their cloud infrastructure be better?


Wait, do people really have a low reputation of Atlassian? The picture o had in mind was that that pay top $$$, are remote friendly, etc. I would have expected their stuff to be much better...


It’s loved by managers (the ones deciding to buy the product) because it looks shiny and it can auto generate burndown charts.

For anyone actually trying to work with it on a daily basis is a huge pain in the ass. Mostly due to the dreadful performance, which was bed enough with the on prem solutions and has now gotten even worse on cloud. Another wtf example is their wysiwyg editor actually has no own markup, it is stored as plain html and has tons of bugs, such that things you edit will return in a different format after you press save, with no way to restore it. This is just so fundamental input form 101 that you wonder what else is equally buggy underneath, especially this being one of their main values. Every one of their product uses a different such editor, all with their own set of different bugs.

Otherwise, you can’t deny that their products are actually very powerful. If only they cared more about basic usability they could have a really good offering.


Not to mention their own version of “not quite markdown” markdown. It infuriates me to try and format text in any atlassian product because of that. I don’t know a single technical non manager person who enjoys or likes atlassian anything.


Haha that reminds me, there is/was a sort of markup language they used for issue comments etc. I had to spit that out of a bug migration tool a long time ago. Back then there was no documentation for it, but I think it was something like this:

https://jira.atlassian.com/secure/WikiRendererHelpAction.jsp...

Their API was unbelievably slow but still better than having to interact with the web app. Please Oh Deities, let me be on an hourly contract the next time I have to use their products!


Yes, the wysiwyg editor is terribly inconsistent between their products. Another thing that is super inconsistent is whether your work is saved as draft or not.

My co-worker once lost 2 hours of work spent on drafting a new JIRA issue (or maybe it was a comment, I don't remember).

There's a feature request in their tracking system for preserving un-saved work. It's been there for almost 20 years now: https://jira.atlassian.com/browse/JRASERVER-4233


Gosh the wysiwyg editor is so, so, so bad. I still can’t believe it’s so bad.


it's not consistent between confluence and jira. they don't listen to requests for improvements (best example is ctrl+[ / ctrl+] for indent / outdent).

the latter is forgivable, the former is not.


Burnnnndownnnnn charttttss!!!

The killer feature, the reason I’ve painfully endured JIRA for about 15 years…

It’s always cited as important but I’ve never actually seen them used in anger.


Performance is terrible in their cloud offering, but actually pretty OK in the on prem version.

Guess which one of those two they're sunsetting...


Yeah, they don't have a good reputation. Not as bad as Oracle (eg "literally evil"), but probably half way there. :)

Their software is "enterprise software", in all the bad ways. eg expensive, slow performance, doesn't work properly, products that don't seem to receive updates after their creating company was acquired (eg StatusPage), etc.

But they have a marketing/PR department, which covers over a lot of those problems.


No they suck. Like really really suck. JIRA is shitty, slow and unintuitive. Confluence somehow makes me hate writing any documentation (which I generally like doing).


People working remotely can write very slow and complex bug trackers too, you know...


I’ve used Jira and Confluence on-prem for many years. They’re fine, although the performance is a bit annoying and the WYSIWYG editors can get confused sometimes.

I actually like Confluence, it’s a well done wiki solution.


Consider yourself fortunate, you were spared pain of using JIRA and Confluence


IME Jira is fine when on-prem and managed by a good admin, but that's become impossible as they're forcing customers to move to their cloud offering.


Our company moved to Jira from Siebel by Oracle - THAT was a true nightmare and so Jira was a huge relief to us. But yes, it was on-prem and managed by a good admin.


Our on-prem sucks too…


We use Confluence (on-premise).

It's fine. Certainly no worse than any other wiki that just accretes content.

(Then again, we're not doing anything overly sophisticated with it, or relying on any of its features for critical workflows.)


It is impossible to find anything in Confluence. The search is useless. Anything that's not in my list of recently worked-on/viewed is basically gone forever.


Frankly, that's no better or worst than any other wiki I've used. It's certainly not something I blame Confluence for particularly.


I can search for strings that exactly exist in pages and it still doesn't find them, it has no way I can find to limit searches to pages I've modified, and its search result ordering is awful, partial search results crowd out full matches. I'm sure there's worse search features, but this is supposed to be decent, well-supported software. It's not some crap noname wiki, supported by one guy parttime.


Mumbles noises about a continuous-scraping system (which possibly runs overnight) and chucks everything into a 3rd-party search solution

(If there isn't that much content, now would definitely be the time to do this, so you entrench the capability and get everyone to depend on it to the extent that it must be fixed once it can't keep up. :P)


thank you for saying this -- I thought it was just me!


What are the pains of using Jira and confluence?


Here's one that bothers me.

If you use Jira to track sprints, and you want to view the previous sprint, how would you expect to do it? I don't mean a sprint from 5 years ago, I mean the one from last week. Let's say you want to view stories that didn't get completed last sprint, or you just want to see what happened because you're doing a retrospective. How would you expect to go about this in a mature product for which your organization pays money?

You might say "well, there's probably a button on the current sprint that lets you go back in time to the last sprint. Like, some form of pagination. Or, maybe you can even set a date on a calendar and see the active sprint during that date. That would be cool."

According to Atlassian's support forum, there is "no way to currently do this"[1].

[1] https://community.atlassian.com/t5/Jira-Software-questions/H...

You can approximate it by building a custom query, or going into a reporting dashboard (if you have access). You shouldn't need to.


I'm able to view previous sprints for a board by going to the reports tab on the left sidebar.


There’s something hidden in the sprint reports. One of the different views there let‘s you see the stories from any previous sprint. Not sure whether it reflects the state of the stories as of that sprint, though.


Confluence supported Markdown (or was it HTML? It’s been a long time.). We wrote a lot of automation that would pull comments out of code, format them that way, and load them into Confluence. Voila! Documentation that’s always synced with production code.

And then a point upgrade removed that support in favor of their own rich text format, so in an instant all our tooling died. We were not happy campers.

I have other thoughts about Jira: https://honeypot.net/post/jira-is-a-code-smell/


It wasn't a point upgrade - Confluence 4.0 got rid of markdown in favour of an "XHTML" storage format, then a layer on the editor to autocomplete markdown into the rich-text as you typed.

Personally I preferred the ability to edit pages as markdown, and we toyed with ways to allow users to edit the rich pages in markdown (i.e. a new transformer from the storage format to wiki-markup, instead of the editor format) - but it wasn't possible to not come up with a diverging solution for the transformations here (at the time, given the time we had spent on it).

Whilst we copped a lot of flack from the die-hard Confluence users when this was done, the vast majority of users liked the change. Also from a different angle, this was a really fun time at Atlassian - we had run out of space in the CornX (office at the time) so we had leased the floors above the pub next door (The Dundee Arms), and there were five of us in there, all with out machines crammed around a single table in this tiny room. Good times.


It was a death knell for us. We moved all our documentation into Git, which the devs found more convenient and usable. I think there might’ve been some survival bias: the people who kept using it after the breaking change liked it. The people who dropped it immediately because it ruined their workflows just moved on (or put it in read-only mode).


Oh completely agree regarding the survival bias - I think the over-arching plan was to make it more accessible to more than just dev teams, and it seems to have worked. "Like MS Word" was thrown around a bit.


Confluence content is write only.

Good luck finding it ever again unless you know exactly what you are looking for.


So much so I wrote a app that I used for finding things in my own confluence and then released as a marketplace application.

You can mitigate the issues in the search if you are organise your content well but the search is still deficient in other ways.


Jira requires a person to tailor it to your org. It can be pretty good if you have people who configure it correctly and have a workflow that suits your team/org, but that's kinda the problem: making Jira be good is a full time job. Confluence is just a terrible experience overall. Terrible editing, terrible browsing, terrible search. I've used both at multiple orgs, I've seen Jira be good, I've never seen Confluence be good.


Yeah, same experience here. Confluence is just dogshit: the only positive thing I can say about it is that it's better as a wiki than using Google Drive with Docs, which can memory hole even faster if it's not set up correctly.


Put a sleep(60000) in every request of your own webapp and it will still be more responsive than JIRA.


Terrible UX and performance so it’s extremely frustrating to use.

I’ve used JIRA in 3 companies and it was always the worst piece of software I had to interact with. Now we just use Github issues and their “Projects beta” and it’s a huge improvement. Engineer first workflow rather than optimized for creating some pseudo insightful views for useless managers


Slow, complex (in terms of feature), painful (in terms of UX)---pick any three.


Inconsistent user experience between the products. “Oh this is Jira so I can’t use those formatting options that I use in confluence.”


It's worse than that; even within Jira the editing contexts are completely inconsistent. And the cache behavior is apparently nondeterministic. It boggles my mind that such awful UX is tolerated in software used in so many companies. It's like Atlassian put all its eggs in the "crack the enterprise sales code" basket, met wild success, and that was all they were capable of. If my job required me to work in Confluence and Jira all day I'd change my job or my career.


Rage inducing formatting


Trello is also terrible software. It's basically a web interface for agile post-it notes.

Compared to decent project management software it has almost no project management features.


That's exactly what Trello is though - boards of cards. Do not expect the usual project management features from it.


This is a funny one. People usually give all the rage to Jira because it's so complex, and Trello because it's post-its. I feel like there's no good middle-ground, people are gonna hate one or the other


can you give me a good project management software to use? alternative to Trello.


Shortcut


Haven't used any Altassian stuff for a few years but when I last did it was all self-hosted. IDK how many shops still do that.


From what I understand, Atlassian is pushing hard to move everyone to their cloud offerings. I think the only way to run on-prem stuff now is to use the Enterprise or Data Center offerings, which are stupid expensive.


And here, “stupid expensive” means $42K/yr: https://www.atlassian.com/software/jira/pricing?tab=data-cen...

Yeah, no.


7$/user/month is pretty reasonable IMHO. Of course only for companies with 500 users. But even if you buy the 500 seat package for 100 users it's just 35$/user/month, compared to normal engineering salaries this is absolutely negligible.


It's not reasonable to post your company trade secrets on a third party, internet-accessible server.


How has no one open-sourced a competing product by now? It feels like the universe where Git was never invented and everyone stayed subscribed to Perforce.

Are executives just scared of getting fired for moving off of Azure DevOps and Jira? They’re both garbage pieces of software and everyone who uses them knows it.


I think the reason nobody open sources a competitor is because most people in a position to develop such a thing don’t think the solution to the overarching problems the Atlassian tools supposedly solve is a similar suite of tools.

I’ve learned to translate “We need to buy into Atlassian’s suite” as “our process is broken and we’re blaming the tools.” This has the standard “now you have two problems” punchline. I’ve seen this play out several times at companies of varying sizes. The endgame “fix” is always that one or more people become entirely devoted to wrangling the complexities of the Atlassian stuff, but a dedicated person/team would have fixed the problem with the old tools, too.


As far as Confluence competitors go, in the vicinity of 2010 there was a lovely MediaWiki fork called DekiWiki (later MindTouch Core). It was easy to set up and use, fast, WYSIWYG editing, attachments. Unfortunately Mindtouch (the company) pivoted into- well, I never quite understood what, but it didn't involve developing deki and it became defunct. It would have been very interesting to see where it stacked up today if it was kept alive.


Because Atlassian pulled a Microsoft and made sure everyone had already irreplaceably adopted their products into their workflows before they made this move.


Confluence is unmatched in the niche it is in - if you compare feature for feature many wikis seem comparable; but the amount of addons for Confluence is stupid high, and many do weird little things that are really helpful for business.


Every place I've worked self hosts and my current company still does. While EOL is still aways away for server (2/2024), it's gonna be a dark day when it arrives.


In enterprise timescales, 2/2024 is not as “aways away” as you think.


I’ve seen the outage mentioned here a few times, but maybe that was in the new queue and it never made it to the front page.


they're maybe down for some people, or some of the time, but they're not totally down.

i've definitely been using bitbucket and jira for the last four days without noticing anything wrong.


I do think I saw this in the front page 3 days ago.


If anyone is looking for alternatives, I’ve found Linear.app to be like a breath of fresh air compared to Jira.

All I was looking for after the slugfest of Jira was fast performance and keyboard shortcuts, and so far linear has been able to deliver.


We use Linear, too, and it’s pretty great. Let me put it this way: literally everyone on the engineering team was happy when we moved to it.


Curious how many boards/tickets you have in Linear. Is it ready for primetime or is it still a tool for small to medium companies?


I'm trying a demo of it and it looks like its mainly aimed at small to medium companies. There isn't enough hierarchy.

The top level organization in Linear is a Workspace. As far as I can tell this should be what your entire company falls under. The next and only level of organization under that is a Team. It really needs something in between that.

Teams have their own Issues, Board (singular), Cycles(Sprints), Projects, Ticket workflows. That means a Team must be a very small set of people (<10 people). So for a company with 1000 employees you might have 100 Teams.

Teams are shown in the UI in a flat list in many places. Expanding the list would likely overwhelm your screen.

The Roadmap feature applies to the entire workspace (your company). In medium sized companies you'll have Organizations with many hierarchies below it comprised of dozens of teams. An Org typically has roadmaps of their own that don't necessarily correspond to company level goals. Teams below the Org will also have their own roadmaps that usually fall under the Org's high level roadmap items but again not always.

There also needs to be more company level controls over settings like workflows. A strict company wide workflow for Bugs is common so that all teams have the same expectations around SLAs for fixes.

All that said I think you could make it work with enough filters and naming conventions.

Also holy hell Linear is fast and easy to use. If I was at a startup I would 100% put us on this.


Linear founder/ceo here.

Thanks for detailed feedback and walkthrough! Just to share some context, we do have thousands of customers, ranging from early stage startups to public companies.

I also agree that we should support more hierarchy when it comes to roadmap, teams and even the workspace settings, and it's something we are working on.

> Board (singular)

Board view supported by other views too, like cycles, active issues, projects etc. The "Board" sits in the navigation for now until we can make the view settings robust enough to fully replace it.


Good to hear! Any write ups on how the larger companies organize and manage their Workspace?

Moving us from Jira/Confluence would probably take an act of god, but I could see teams going rogue and doing their internal planning in Linear while still using some part of Jira to ensure they stay connected to the rest of the company process. That way we could maybe do incremental adoption.

Is there any bidirectional Jira <-> Linear support? e.g. Someone creates a bug in Jira for us and it replicates into Linear? When I close it in linear it closes it in Jira.


There is no write ups at least publicly yet. If there are specific questions we can try answer those based on what we have heard from other companies. ks@linear.app

We do have a bidirectional Jira integration. It pulls in the basic information, and then also adds a Jira "attachment" & link to the Jira issue. If you close the issue in Linear, will also close it in Jira [1]

If you enable the Triage feature, it works nicely in a way that requests from other teams in Jira automatically come to the Triage queue and you can review them there. Same triage queue also works for customer support tools, sentry, creating issues through the API etc.

[1]: https://linear.app/docs/jira


Yeah I just found it. The hard map of Jira project to a Team makes it tough. We currently do 1 Org (hundreds of people) = 1 Jira project. Teams are mapped to Jira Components. I think without some level of customization on the sync we wouldn't be able to incrementally adopt. Everyone uses Jira differently unfortunately.


+1: we use them too and they have a reasonably flexible product in terms of labelling and project definitions. The only notable complaint I have is that the search doesn't support negative filters.


Out of curiosity, had you tried Jira Next Gen or only Jira Classic?


GOOD. This company needs to fail, and the world needs to be spared its toxic shitware, aka the wet dream of mediocre micromanagers.


On https://www.atlassian.com/trust/security/data-management, Atlassian claims they regularly test their ability to restore products in less than 6 hours.


They take RDS backups cannot restore them as per:

Can Atlassian’s RDS backups be used to roll back changes? We cannot use our RDS backups to roll back changes. These include changes such as fields overwritten using scripts, or deleted issues, projects, or sites. This is because our data isn’t stored in a single central database. Instead, it is stored across many micro services, which makes rolling back changes a risky process.


Microservices - make every action a distributed transaction!


Class action!


I hate this company and their products so much. The only reason I ever use them is because every single middle manager is obsessed with Jira. Jira sucks and Atlassian is a terrible company.

Also, screw Atlassian for acquiring Trello.


This is more informative: https://confluence.status.atlassian.com

Apparently some portions of their cloud instances have been down for days.

I'm going to really be sad when I can't limp this server instance forward any more.


Interesting. Bitbucket and confluence work for me, but perhaps some users are being load balanced to the servers that are down.


Correct. It's only a small percentage of users.


I can see being down for a couple of hours or possibly even a day. But 4 days is a bit much. Couldn't they restore from a cold backup in less time than that?


JIRA was up and fine on Friday; so, not all of their products have been down that long.


According to sibling comments not all users are affected alike.


Maybe not if they didn't have a tested recovery plan for whatever failure scenario they are experiencing.


I don't have much context but I assume given it only affects a small subset of customers, they can't restore from a cold backup. Since the unaffected customers would then experience data loss.


I had pleasure of using HipChat some times ago. That was probably the worst Atlassian product I have ever used. You want to edit the message you just send, you have to type 's/string/replacement' within 60s.


Thank God. I hope it dies off. It is such a terrible bloated product. I spent more time waiting for it to load when I had to use it than getting actual work done.


Is JIRA really globally unavailable for 4 days? That's simply incredible.


It seems to be partitioned by account. The Jira/Confluence/Opsgenie accounts that I use have been gone since Tuesday, but it's clear that other sites/accounts are working fine.


I have have been using it all week - up to and including yesterday. No issues.


I’ve been using JIRA in the last four days without any issues (other than a potential issue with the iOS app, but I was in a poor data area at the time).


Ironic how their "Statuspage" product is having issues.


Happened while they team having fun with the event

https://events.atlassian.com/team22


Don't most people self host an instance still?

I haven't used it in a few years.


Are you dismissing the importance of points driven scrums? And what about the stories? Won't someone think of the stories?!?


As the days turned into years, and the issues into subscriptions; verily did the stories themselves turn to horror stories, and finally into full horror novels, from which no sprinter could ever escape.


I'm so glad it's been years since I've had to that nonsense


Did consulting work for a place that gave their execs bonuses based on story-point velocity... jesus.


"What else we can do to improve productiviy... except paying more to whose who actually do the work" ?


They have EOL'ed most of their self hosting products. I don't think you can buy a new self host license anymore.


Even if you could, you wouldn't. Self-hosted will lose maintenance support and you'll need to then move to the cloud anyway.


They've changed the on-prem pricing significantly. If we were to have renewed, it would have cost us 70x more this year.


We just transitioned to Bitbucket, and I've been using confluence and jira all week (including today) with no issues. There may have been outages and latency, but it's not been "down" for 4 days.


It seems the outages have been isolated to specific instances. Not everyone is having problems but a significant amount of customers have been "down" for multiple days.


Why? It's been years since I've heard anyone transition to Bitbucket. Self-hosted or SaaS, even companies that have fully embraced Atlassian tools don't move to Bitbucket. There are much better options out there, in terms of features, speed and reliability.


None of the product owners can put a bug ticket in so they have no way to proceed.


Does anyone know why their stuff has been down for four days?


Rumour on the streets of Sydney is that the test instances and prod instances got mixed up in a database migration.


Wow really? Was the rumor that they overwrote a prod db with a test db?


No but if you read the updates on their status pages it really sounds like they are making it up as they go. Test restorations, working with small groups of customers to verify recovery, etc.


You would expect that, damage limitation. If rolling out a fix you would want to roll it out to your smallest customers first to verify then incrementally roll it out to more/larger customers as your confidence increases. Especially if it’s anything data/db migration related as some comments hint at and the mention of possible data loss being minimum in one of their official responses suggesting something data related.

On the status updates I guess they can’t say more other than they are working on it. Saying more at this time might open them up to other issues considering a lot of businesses are run off moving Jira tickets around today.


I remember that time a few years back when Bitbucket was down for days after they somehow dropped their entire storage layer. We never received a postmortem of any kind.


Used Jira many years ago, always was unimpressed by it. A multibillion dollar company and most of their core products are down?


They recently redid some of their core infrastructure so that may have come undone.


Ah, refactoring strikes again!


There's never time to do it right, always time to do it over.


Shamelessly using this to promote my product: https://recalllab.com

It's build with the vision of what I'd want a project management tool to be. Simple to use, useful to help manage between teammates and ultimately reducing the time I need to spend on it.


The "New GitHub Issues" (beta) is very actively being worked on, with monthly feature updates and blog articles and everything. Highly recommend to try the beta.

https://github.com/features/issues


I guess they better increase the difficulty of their coding interviews to LC hard to fix this problem.


When you employ dark patterns, that's what happen. I hope they fail miserably.


Fantastic news. Hope they're using Jira for the fix ticket


"Just let us host it for you," they said...


Anyone know if there's a cloud export option? I'm not an org administrator, but if I were I'd want to keep an offsite backup of our data. I could throw together a script to do it (their APIs are easy to use and their Python module covers pretty much everything) but obvs would be great if you could just do a bulk export.


Yes. You should routinely generate and pull down a backup. We do it once a month.


No love for Redmine ? https://redmine.org

* Ticket tracker

* Email workflows

* Project management

* Wiki

* Free open source


I don't quite understand this. I've been using Jira and Confluence all day. What's the problem?


Not all customers are affected (sounds like a very small percentage are), but those affected have been unable to use their Jira, Confluence, and Opsgenie for several days.


Could it be related to this? https://mobile.twitter.com/Gi7w0rm/status/151278985519709389...

Further down they mention Atlassian is affected. Unsure what to make of it


So when are we going to contact the CEO of Atlassian, as all of their products and services have gone down and taken an unexpected 4 day holiday into next week?

Would be a disaster if come Monday, we also have parts of GitHub's service deciding to go down for an unexpected holiday as well.


We switched from Jira and started using Zoho Sprints. It is a great tool for agile teams and their pricing is much lesser compared to other alternatives. Their reports are really great and the product is growing really fast with new updates every week.


Hosted Jira was working fine for me yesterday.


4 days is kind of insane and unacceptable


I've never seen a downtime like this, and I've never seen a company like this.



To be fair, developers everywhere are probably more productive with JIRA unavailable.


But how will managers feel useful?


Now to mention, scrum "coaches" and that entire industry?


LOL no. If you think this way, there's something wrong in which your company is using it. It helps me personally to manage my work better.


It’s a near uniform “joke” at this point.

Eventually you need to stop blaming the user.


It's not always the end user's fault, often its the organization's maturity, or lack thereof, as far as tasking.

However, pretending that there isn't a mass of programmers who think they know better than everyone else and couldn't benefit from task structure / documentation (be it SCRUM, Agile, whatever) isn't helpful at all.


What issues do people have with Jira? I've always wondered.


It absolutely has some nice features like how Jira and Confluence and BB can all work together.

But they prioritize feature velocity towards being "all things to all people" over usability. This means it's a mountain of inconsistency, bloat, and instability. It's super hard to use if page loads are many seconds long, your favorite button got moved into a hamburger menu to maximize unused whitespace, and their servers are a tangle they can't manage.


Speaking of consistency, IIRC the markup languages are inconsistent between Jira and Confluence. Why would you do that?


The most recent annoyance for me is the fact that you can't use it as a simple issue tracking system.

Without firing up my work laptop I can't provide a precise number, but when you want to file a new ticket, there are at least a dozen options for a ticket type, and once you've created a ticket you're stuck with the consequences of your selection.

Just give me a single ticket type. Let me create any relationship I want between tickets. If I want to assign one ticket as parent for another one, I should be able to do that.


I forgot one of my frustrations: when it’s time to create a ticket, the board selection drop down doesn’t display the full board names, so there’s a few seconds of friction while I try to remember which board I want based on a truncated name.

Or, often as not, I create it on the wrong board and can’t move it.


It went from an highly efficient and customisable ticket/work tracker to a clusterfsck of a feature creep mess that requires 10 clicks to add an item to a custom field, that’s when you remember where to look because if you are not an atlassian PhD, it will take more time to search how to do something than simply implementing it.

  I am still using it because my team is used to it and I don’t have the time to migrate the 3gb+ Database somewhere else for now.
Actually this is a good point, anyone migrated somewhere else using the backup? What was your experience?


The XML backup that is available in the Server and Datacenter editions was not meant to be used for anything other than toy instances. The documentation suggested that you restore from database backups and copies of the home directory that contained all the attachments.

Whenever possible, I've always used SQL backups (MariaDB or PostgreSQL) to clone or restore JIRA instances. The database information will accept changes in the base URL without issues. It makes setting up testing and mimic environments pretty easy.

When I recently had to move a project to a new instance that I didn't have jira-administrator access to, I had to provide the server admins a set of CSV exports of our JIRA project so that it could be imported over properly. All of the tasks subtasks, and linking had to be provided in separate CSV files because JIRA would fail to export more than 20000 entries in a single file.

It took weeks worth dry runs and checkouts to make sure it would be successful. This was because we had to give this other team a long list of custom fields, workflows, and schemes that represented our software development and program management methodologies.

On the transition date, I was out of the office on a prearranged cancer treatment. I still got texts from people freaking out while I was sitting in the cancer ward getting my infusion. I told them that I couldn't even look it anything until I got back in the office after my treatment and recovery period. Thankfully, they managed to limp along get 80% of the issues resolved within 2-3 days.


How about the fact that it’s artificially slower if you’re not paying for the higher plan? For starters.


Can you provide a source for this claim? Haven’t heard of this before.


The slowest UI I've literally ever experienced.

Any action you take has a 20-40% chance of not actually committing.

History view in some parts of it is literally useless.


Just avoid the ui and use the CLI instead. https://github.com/go-jira/jira Granted getting your custom templates configured with your custom fields set up by the administrator can be a giant pain.


Alternatively just say no to JIRA.

JIRA is one of the main reasons I left let's encrypt. JIRA is also one of the main reasons I said no to two companies.

Life is way too fucking short to deal with this. I think companies don't realize how much devex matters. A big reason of employee attrition in my experience has been shitty fucking tools that never get replaced.


This is good information. I used it 5 years ago and my memories are not particularly bad. But it was self-hosted by a capable admin and the people customizing it were decent developers.

From what I read here things have changed a lot: They force everybody to cloud which is slow.

And the fact that they are higly customizable (adding mandatory fields...) is of course a nightmare in all companies with too many incapable managers, which is probably a fair share...

I use gitlab these days. Its issue management has limitations, but being mostly snappy (despite cloud) and lightweight is a better exeperience than the nightmare I am reading here. I can run my own REST queries if some limitation bothers me too much.


Yeah there's just so many better solutions. Unfortunately a lot of companies think this is a low priority issue and never actually spend time replacing it.

Everyone at my workplace hated jira. We still never replaced it.


I’ve used it at multiple companies. I have nothing against it.

But the experience of using it is 100% in the administrator’s control. It works quite well in the default form, but it’s HEAVILY customizable. And that seems to be where most problems come from.

Trying to cram workflows where they don’t fit. Adding tons of extra required fields. Making special rules about how you have to use it.

I’ve certainly seen it go from good to annoying from new people deciding to “fix” some process by adding to it. Including the day you couldn’t create tickets because our admin added a required field that was hidden so you couldn’t fill it out.


It can be good if configured well, but it’s almost impossible to configure well. I don’t even know how to describe it, it’s so difficult to change settings to make things better and not worse. One can make different issue types, and each issue type has fields and workflows, and you can make different screens to deal with all of those things. It’s just so easy to add complexity and so difficult to remove it. Addons often add all these things, and trying 3 addons for 1 hour just to compare them can really mess things up if you don’t know what you’re doing. It’s even quite difficult if not impossible to do this in a completely isolated test environment. And of course you’ve often got people adding complexity on purpose.

Once complexity is added, it’s basically impossible to to reduce it, even by starting a new project with new screens, etc. Because at minimum your issue types will all be infected with complexity. The most competent managers I have seen try JIRA for 2-3 weeks max and then switch to a spreadsheet.


Jira is famously customizable, and middle managers famously add billions of fields to the bug pages, resulting in incredibly slow page loads.


That: middle managers famously add billions of fields to the bug pages - sounds strange. Jira should reflect a software dev process in your organization, it just helps you to follow it. The process is defined by a doc which can be changed, but not on a whim of any middle manager.

Your description implies that your process is broken, but if your process is broken no Jira can help you. I doubt than even God can help you…


Exactly, and many places have broken processes that are revealed by Jira. Before the dam holding back process stupidity was "the system can't do it", but Jira can do it.

Unfortunately once process stupidity sets in it usually metastasizes and there's no cure.


The mains complaints I see people make are that it's slow (which usually gets the response that it's configured badly), and something I'd summarize as installations being configured based on what management thinks they want rather than what the devs "know" they want. Which latter point would of course apply to any flexible enough tool used under the same culture.


But it's where you get work, and sometimes people put shit in it!


Yeah I hear this often enough that maybe its just too hard to setup correctly… but no it’s the users who are wrong.


But, of course, we will never admit that Java is a fucking abomination and designed by brain dead degenerates.

And yes, I studied programming languages before coming to this particular conclusion.


I did not study programming languages and also came to that conclusion




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: