GitHub Outage Map

Community Discussion

Tips? Frustrations? Share them here. Useful comments include a description of the problem, city and postal code.

Beware of "support numbers" or "recovery" accounts that might be posted below. Make sure to report and downvote those comments. Avoid posting your personal information.

GitHub Issues Reports

Latest outage, problems and issue reports in social media:

• 
  Sage (@HelloSage_) reported 3 minutes ago

  @TheHackersNews Shadow AI is one risk. The other is what happens when those agents process untrusted GitHub issues. A new benchmark had 66.5% of malicious ones sail through the guardrails on Cursor, Claude Code and Codex.

• 
  TheMakerOfMedia (@TheMakerOfMedia) reported 3 minutes ago

  @catgirlprostate I get what you mean but saying this in response to github is just out of touch lmao. it even isn't a github issue some pages have a very apparent download section while others just hide it in the crevices. some coders just dont deserve to enter the thousand year kingdom

• 
  Felix Fong (@_fuccii) reported 6 minutes ago

  Github stacked PRs have a dangerous bug: CI only runs against the contents of the top most PR in the stack. A bottom PR can be broken but a top PR with unrelated changes can still make the entire stack appear green.

• 
  Tom Hosiawa (@thosiawa) reported 7 minutes ago

  @bentlegen I have, here's one where it wants to push a fix upstream without it even asking me. In a previous case, I only realized it after it said it was blocked by github after it tried I think we'll need to start adding "refuse this rationalization" scope checks

• 
  Shai (Deshe) Wyborski (@DesheShai) reported 9 minutes ago

  Thanks for the new "paper" David, I could do with a good laugh. But I was talking about math, not about writing a bunch of mathy-looking stuff and posting it on your GitHub for your little fans to gush over. Full disclosure: I didn't read the whole thing. I didn't have to. I only got up to page 2 before running into a blatantly fatal error, and in page 4 I found another one, so I didn't see a reason to push forward. Maybe that's why you had to present it in workshop instead of trying to get it published. Anyway, let's get to it. In page 2, you said “Compensating for this decline […] requires the attacker to add a supplementary transaction containing real fees.” That's false. It does not follow from requirements 1-5. This assumptions hide the fact that your "analysis" implicitly assumes that attackers always try to preserve the same-production work. That is, your "proof" only prohibits one attack vector, and disregards most realistic scenarios such as: - Routers that insert Sybils without producing the block - Attackers whose block already contains enough work Explicit example: without Sybiling at n=H=2, the atataacker routing share is u(2,2) = 1/3. After adding one identity, we have u(2,3) + u(3,3) = 2/7 + 1/7 > 1/3 so the attacker controls hops 2-3. That is, whenever this transcation is included (which has unchanged positive probability. E.g. if the eventual block producer already satisfies the work threshold from other transactions, inserting this additional hop need not change the transaction’s probability of inclusion), Sybilling strictly increases routing payout for free. That's not the only gap I found. You derived propagation asa preferable only when x > y/7, but then kinda just assumed this means y>7x is "impossible". More egregiously, you apply the inclusion-exclusion principle while switching between the distributions x and y. These distributions might have the same sample space, but a different probability measure induced by different strategy. Hence, using inclusion-exclusion to establish a relationship between them is a rookie mistake which completely invalidates your proof strategy. This invalidates your entire proof strategy. Applying the exclusion-inclusion principle while using a different probability function for each event is a rookie mistake. Your paper might prove that "compensated Sybil" is a bad strategy (which is not exactly a surprise. Why would anyone want to do that?). But it's a far cry from "disproving Babaioff" and is exactly what happens when cryptobros think they are mathematicians. They do a lousy job all around, and post it to their GitHubs so that their little following of people who think they are going to make them rich will laud how smart they sound to the uninitiated. Either way, having found an actual deep flaw in your "work" (thanks for bringing it up!), I'll mail my concerns to IWGTEA, which might save them some embarrassment. But again, thanks for the laugh. I needed it.

• 
  Steve Avon (@SteveAvon40) reported 13 minutes ago

  Made a Code Review AI Agent for @salesforce - Client uses Copado for ticket management and Github for source control. After a ticket is committed and documented, developer can mark a ticket as ready for review @agentforce agent will then fetch the components from the correct branch in @github, and compare them against our component standard documents, which are stored in SF Knowledge. If there are any issues with components, or the ticket in general, chatter post will be made on the record and promotion will be blocked until everything is corrected and resubmitted. Been pretty fun to work on and saving a lot of time for the code review team.

• 
  Magnum (@AltHarbor) reported 17 minutes ago

  @MrKingLollipop @cursor_ai @OpenAI I misunderstood. Good question. I’m guessing they haven’t updated the pricing this quickly. I’m a sicko a bought 2nd Claude sub. Cursor is so good and fast for a lot of stuff but only Codex and Claude can handle the long running huge workflows I need right now with the tools they offer. The API subsidy in cursor doesn’t get me far enough for what I get out of the Codex and Claude subs. Honestly bugbot reviews on GitHub actions is my most valuable use for it currently. Problem is it all gets billed through API usage instead of the standard usage which is lame. Biggest complaint by far.

• 
  Stefan (@schteppe) reported 17 minutes ago

  @dimabytes @madebygps Because the CI build is slow. Sometimes 30 minutes. During that wait, the dev can continue building on top of the previous PR changes. (We’re not using GitHub specifically but we implement stacks in a different way)

• 
  Steven Isaac (@onchainsteve) reported 19 minutes ago

  A Year of Work on SolOnChain, Ended in a Discord Message Yesterday at 17:35, a year of work ended in a paragraph. No call beforehand. No "we need to talk." Just a message saying they'd decided to move to a new development team, that it made the most sense financially, and that they wouldn't be needing us on the platform anymore. Thanks for everything, best of luck. I'd spoken to them two days earlier. Nothing came up. I'm writing this because I've been a developer in this space long enough to know how common this is, and how rarely anyone says it out loud. What we built Myself and my co-dev at Bytez 3 built SolOnChain, the NFT launchpad and marketplace on Solana. The launchpad, the marketplace, the infrastructure underneath it. We were paid for the initial build, and that was the honest scope of the agreement. We didn't stop there. We kept adding, kept fixing, kept shipping features that were never on any invoice. Not because we were naive about it. We'd discussed a long-term partnership, and we were building toward that. When you think you're in something for the long haul, you don't nickel-and-dime the roadmap. You make the product good and you trust the relationship to make it worth it. That's the part I want other devs to sit with. The extra work wasn't charity. It was an investment in an arrangement we thought we had. The wind-down When the engagement ended, we did what you do: we disconnected our GitHub account from the SolOnChain Railway project and removed the DDoS protection that had been running through our licensed enterprise account. That protection was ours, paid for by us, and it isn't something we can leave running for a client we no longer work with. Their new team will need to set it up themselves, and we told them so directly rather than letting them find out when something fell over. We also offered a dated handover document confirming that any future updates to SolOnChain aren't affiliated with Bytes 3. That matters for both sides. If someone else is maintaining it now, our name shouldn't be attached to what happens next. I want to be clear that we haven't torched anything. There was a moment yesterday where going open source with the codebase got mentioned in frustration, and we thought better of it. That's not who we are and it's not how you end things. The part that actually stings It was never really about the money. We'd had the financial conversations before, more than once, and we'd been understanding about them. We'd had fall-outs too, over money, over stress, over the ordinary friction of a project this size. That's normal. Anyone who's shipped something real has had those weeks. What I didn't expect was to find out we were finished by reading it, at the same time as everyone else, with no conversation first. A five-minute call beforehand would have cost nothing. It would have let us wind down properly, hand over cleanly, and part on genuinely good terms. Instead the first thing anyone said after the announcement was "wait, where has this come from?" That's the whole complaint. Not the decision. That was their business and their call. The way it was delivered. What I'd tell any dev reading this Put the partnership in writing, not just in the conversation. A long-term arrangement everyone verbally agreed to is worth exactly nothing on the day someone changes their mind. If it's real, it's on paper. If it's not on paper, price your work as though the relationship ends tomorrow, because it might. Settle ownership before you write a line of code. Who owns the repo, the deployments, the domains, the credentials. Not because you expect a fight, but because the absence of an answer is the fight. Every one of these disputes I've seen comes down to two parties who each sincerely believed they owned the thing. Separate your infrastructure from theirs from day one. If your licensed accounts are load-bearing for their production environment, you've created a mess that will land on you at exactly the worst moment. Bill the extra work, or accept it as a gift. Those are the only two honest options. "I'll sort it out later on the strength of the relationship" is not a third one, and I say that as someone who chose it. Have the exit conversation early, while everyone still likes each other. Notice periods, handover terms, what happens to outstanding balances. Nobody wants to raise it during the honeymoon. Raise it anyway. Where I'm at I'm proud of what we built. That doesn't change. SolOnChain is a strong platform, we're better engineers than we were twelve months ago, and I'd back that codebase against anything comparable in the space. I hope it does well. I mean that. I've named the project because it's the project, and anyone who's used it deserves to know who was behind it. I haven't named individuals, and I'm not going to. This is about how a business decision was handled, not about people. I'm not writing this to burn a bridge. I'm writing it because devs in this industry talk endlessly about launches and almost never about this: the ordinary, unglamorous way a year of work can end in an afternoon with no conversation attached to it. It happens constantly. It should be said out loud more often. To anyone in the SolOnChain community who enjoyed using something we made: thank you. That was the point. That was always the point. And if you've had this happen to you: you're not being precious for feeling it. You built something. Of course it lands hard. Bytes 3

• 
  Jeff Lindsay (@progrium) reported 20 minutes ago

  the next github should have project member status. so project members can say what theyre doing or if out of town or will be unresponsive. "heads down on next version", or "reviewing PRs after conf next week" ... hell, prob good for agents too

• 
  Dmytro Bavykin (@DmytroBavykin) reported 28 minutes ago

  Can Fable 5 fix your CI/CD pipeline at GitHub? How caching could speed up your delivery? Those were the questions I asked myself when Fable was released. What I knew for sure: - there were build steps taking more time then they supposed too - test suite was split in shards - tests were taking long - cache thresholds were over the limit I did not expect Fable to fix it all in one go, but I did expect it to be capable to analyse those pain points. So, I fed to it the folder with pipeline configs, different workflows, pointed to GitHub cache storage, etc. The results were shockingly good. It found a long list of improvements, but you shouldn't take them on faith. Even with a model at Fable's level, you need to make it doubt itself. Some suggestions were contradictory or overlapping and a few were outright redundant or over-engineered. After you have something to start from, ask your research LLM to note it down in .MD guide for another model to execute it. You don't need top tier model for changing lines in YML-files or for making an API request to GitHub for measuring Cache Storage size. So I tasked Fable to make a guide for Opus. Before I go into more details on what was fixed, so you can run the list against your pipeline too, lemme flex by saying that CI time went from crazy 1 hour to up to 20 minutes (14 minutes in the best case scenario) in 2 weeks. The biggest thing to fix was caching and specifically how it kept hitting the storage limit. What I found was that too many places were generating cache instead of reusing what already existed. The result was storage that ran hot on writes and cold on reads. Also, when the storage hits its limits, GitHub runs its LRU eviction policy which means that even a fresh cache created in your working branch can be evicted by someone else's. Back then, the oldest cache entry was 20 minutes old. Now it's 6 weeks old, and it was last used 7 minutes ago. The cache is inherited from the parent branches. You may want to have an async job for warming up the cache for your main/develop branches, so its child branches would inherit the cache and save the compute time. One last note, on staleness: by default GitHub removes caches that haven't been accessed for 7 days, but you may want to clean up sooner depending on your workflow. Why it worked? - Container pre-build stage went down from 8 minutes to 1 minute 20 seconds - Some routines that were running sequentially now run in parallel - Then tests were split into shards with installing additional dependencies (each was taking ~2 minutes) - Database seeding ~2 minutes (warm up job now runs the seeds, makes a dump and restores it per shard, take ~2s) Part of it was running in each shard, so you can multiple by 10 shards. One CI run is now 200 compute minutes cheaper. The bill gets smaller, the code reaches the end users faster, the developers are unblocked with shipping new features.

• 
  Nazar Mokrynskyi (@nazarpc) reported 33 minutes ago

  This happens surprisingly often as I try new things. The big question is: how are some developers work for years without a single public issue or pull request on GitHub. What are they doing so that they don't hit anything worth reporting for years? Seems literally impossible.

• 
  Tony Javier Sanchez (@MacroSinFiltro) reported 33 minutes ago

  @github Stacked PRs solve a real human problem: reviewers can reason about 300 lines, not 3,000. The hard part is keeping the stack current when main moves and lower branches change. If GitHub nails rebasing and CI context, this could meaningfully improve review quality.

• 
  Mephistopheles (@mikergmoj96151) reported 34 minutes ago

  @AnimeSerbia the problem is that programmers insist on using github as the main hosting service for their **** and drive normalfags to github. it's a self-inflicted wound. jon blow is right, open source people literally don't understand human behaviour.

• 
  Sebastian Buzdugan (@sebuzdugan) reported 34 minutes ago

  @aubymori github guides help, but permission errors are where beginners still get stuck