dockerdash is in @kestra_io gallery now

watches your Dockerfile, checks Docker Hub for newer tags, opens one bump PR

5 merged, 2 in review

winter arc is in full swing

screenshots below, links in the comments

an http 200 doesn't always mean an application is healthy

it proves an endpoint answered, but a dependency could still be broken

health checks should match what you're actually trying to protect

what makes you trust an unfamiliar open source repo enough to run it locally ?

a clear README, recent maintenance, tests, pinned dependencies, or something else ?

what's your first check before running someone else's code ?

when a deployment goes wrong,

do you prefer rolling back quickly or investigating first and rolling forward ?

i know context matters,

but what's your default and why ?

terraform engineers:

what do you do when someone changes infrastructure manually outside terraform ?

reconcile the configuration, import resources, or investigate the drift first ?

curious how you handle it on real teams

when a deployment fails halfway through,

what do you want first:

an automatic rollback, a clear failure report, or a pause for human approval ?

what decides the right choice on your team?

when a pipeline fails only in CI,

compare the environment before rewriting the code:

variables, permissions, dependency versions and working directory can differ from your laptop

aws engineers, how do you approach debugging an access denied error ?

do you start with the identity policy, resource policy, permission boundary, or cloudtrail ?

curious about the order experienced engineers follow before changing permissions

a smaller docker image isn't just about saving disk space

fewer unnecessary packages can also mean less software to maintain and a smaller attack surface

image size matters, but what you put inside the image matters more💯

docker health checks tell you whether a check passes,

not necessarily whether users can complete the task they came for

what would you check beyond process health?

the interesting part of automation isn't triggering a workflow

it's deciding what should happen when the workflow itself fails

because now you have a system that needs to recover from its own failure

my debugging rule:

change one thing, reproduce the issue, inspect the result

if you change five things at once, you might fix the bug without understanding it

and that makes the next failure harder to solve

a restart can make an application look healthy again without explaining why it failed ?

recovery and root-cause analysis are two different problems

good automation shouldn't confuse one for the other

when a kubernetes deployment starts failing,

what's the first thing you check?

events, pod logs, resource usage, or something else?

devops people, when an alert fires,

what do you want first ?

the raw logs, a dashboard link, or a short summary of what changed?

everyone's talking about the Roadster thrusters

the fan system is the more interesting engineering problem to me

it lowers pressure under the car to generate downforce at low speeds

but how do you maintain that pressure difference as ride height and airflow change?

@MKBHD

a container running doesn't mean your app is healthy

check whether it can actually do its job, not just whether the process is alive

not here to collect tools for my resume

i want to understand,

how code becomes containers

how containers become deployments

how infrastructure recovers when things fail

still learning still shipping

still breaking things so prod doesn't have to

one thing cloud projects taught me the hard way

creating an ec2 instance is easy,

building the network, securing access, managing terraform state, adding monitoring and making the whole thing reproducible is where the actual engineering begins

i used to think automation meant saving a few clicks

now i'm building workflows that detect failures send alerts create issues and recover when things go wrong

the real goal is to build systems that don't need babysitting 💯