@brian_scanlan Well written “we have strong hunches and puffs of smoke, but no definite cause” report. Those are the hardest!
English
Garry Shutler 👨💻📅🚴♂️
11.8K posts
Garry Shutler 👨💻📅🚴♂️
@gshutler
CTO and co-founder of @Cronofy - rickroll enthusiast
Katılım Ocak 2008
839 Takip Edilen954 Takipçiler
Wrote another status page today: intercomstatus.com/us-hosting/inc…
English
@nateberkopec CPU tends to be a leading indicator of queue latency. We scale on both, with CPU being the active one most of the time, but queue depth there as a safeguard.
It usually only kicks in if an external API is having a bad time, and so the IO/CPU balance shifts.
English
Autoscaling based on CPU for background jobs? Might as well not even have an autoscaler.
Imagine you have a job that waits on an external HTTP call for 1 second. Enqueue a million of them. What's your CPU usage burning through this stack of jobs? Does your autoscaler trigger?
English
@jeremysmithco @domster github.com/gshutler/dotfi…
Fuzzy search cd for repos in my Code directory helps with this.
English
@domster Yeah, TIL about the Go way! :)
Yeah, that's a good point. For less common repos, it can be a challenge to remember the user/org name.
English
I have a Code folder under my home dir for all my repos, as well as any others I clone. It's gotten unwieldy, so now I'm using subdirectories by GitHub user/org name (e.g. ~/Code/rails/propshaft). Anyone else do this?
English
More. More! Join us! gshutler.com/2024/06/no-kno…
Tuomas Artman@artman
A few months ago, we changed the way we address bugs @linear. We prioritized bugs over everything else. If you have bugs assigned when you wake up in the morning, you don't do anything else before they are addressed. This approach felt scary and quite radical, but our theory was this: Every software product has defects and the amount of defects being found is somewhat constant. We want to build a product that is of the highest quality, so eventually all bugs will be addressed. This means that the amount of work required to tackle bugs is the same regardless of whether you address the immediately or whether you prioritize feature work and tackle bugs in bulk between feature work. And this made it obvious that the better alternative is to prioritize bugs over everything else. The cost for doing this was that we had to spend a month or so to bring down the backlog of bugs, but once the backlog was eradicated, there was no additional cost to this approach. Now, if we or a customer finds a bugs, it usually gets fixed the very next day. You certainly should consider doing this at your software company.
English
@igor_alexandrov And then I’d make it:
return if current_user
redirect_to login_path
English
@igor_alexandrov Trick question, none stops line 5 running when there’s not a current user resulting in a NoMethodError.
The guard should be a before_action stead.
English
@asmartbear @davidfuesser I like this most from your various replies.
Resonates most with me. I have strong opinions but I’m happy, delighted even, to be proven wrong. Open mind is a nice, short version of that.
English
The problem with the phrase “strong opinions weakly held” is that if they’re weakly held, they not strong.
Maybe it should be:
> Clear opinions weakly held.
Or:
> Opinions held with a strength proportional to evidence.
Or what should it be?
English
@bitfield I hope so, because then we can quietly get on with building like we have with Ruby for the past 5-10 years 😂
English
I'm increasingly getting the sense that Go has passed its peak of popularity. What do you think?
English
Garry Shutler 👨💻📅🚴♂️ retweetledi
@dwedigital I reckon people actually read the post the third time I mention it 😆
English
@dwedigital Thanks Dave! I've referenced it at least twice this week myself! 😁
English
@grumpyjames Ok, errm. Would be quite a capitulation now.
English
Would've preferred that to go the other way from a keeping things exciting perspective, but tomorrow is another day
English
@Mark_Kellett Triggering workflows like “deploy this to nonprod”, “expedite to prod”, etc, based upon comments and such like.
Utility type stuff it would be good to define once and use across a few repositories.
English
Creating a collection of GitHub Actions that can be used by other repositories in the same organization is weirdly difficult.
It’s taken a good hour or so of faffing to get a bash script to run when this feels like something that should have first class support.
English
@GergelyOrosz @donkersgood The difference is that AWS had to buy the hardware to avoid the FOMO and need a return.
Most other business can PAYG something like Bedrock and their only loss is time to work on other things.
AWS must be extremely deep in the hole for this business stream.
English
Such a good example of the GenAI FOMO eating up many tech businesses.
It’s about AWS and how AWS is *so* obsessed with GenAI that it seems to have entirely forgotten about its core ($90B!!) business.
If AWS does this, no wonder so many other businesses do? Via @donkersgood
English
What's the most important lesson you've learned in your career as a software developer?
English
@lawrjones We only use Kubernetes features if we really need them rather than using every feature we have an excuse to.
Less complexity, less to break, less to reason about.
English
@lawrjones Similar experience. Aside from upgrading to the latest version every 6 months or so we don’t really think about it.
There’s also a few times it’s made things crazy trivial compared to the alternatives.
Think a lot comes from our approach of KISS.
English
Everyone says Kubernetes is the worst mistake you can make, you’ll lose so much time debugging it, etc.
Our 2 person infra team runs k8s for our 25 person eng team. Deploy 15x a day, all automated, super smooth.
It is so low drama the infra roadmap has never mentioned it.
YMMV
English