• 2 Posts
Joined 1 year ago
Cake day: July 31st, 2023


  • Some troubleshooting thoughts:

    What do you mean when you say SSH is “down”:

    1. connection refused (fail2ban’s activity could result in a connection refused, but a VPN should have avoided that problem, as you said)
    2. connection timeout. probably a failure at the port forwarding level.
    3. connection succeeded but closed; this can happen for a few reasons, such as the system is in an early boot up state. there’s usually a message in this case.
    4. connection succeeded but auth rejected. this can happen if your os failed to boot but came up in a fallback state of some kind.

    Knowing which one of these it is can give you a lot more information about what’s wrong:

    System can’t get past initial boot = Maybe your NAS is unplugged? Maybe your home DNS cache is down?

    Connection refused = either fail2ban or possibly your home IP has moved and you’re trying to connect to somebody else’s computer? (nginx is very popular after all, it’s not impossible somebody else at your ISP has it running). This can also be a port forwarding failure = something’s wrong with your router.

    Connection succeeded + closed is similar to “can’t get past initial boot”

    Auth rejected might give you a fallback option if you can figure out a default username/password, although you should hope that’s not the case because it means anyone else can also get in when your system is in fallback.

    Very few of these things are actually fixable remotely, btw. I suggest having your sister unplug everything related to your setup, one device at a time. Internet router, raspberry pi, NAS, your VM host, etc. Make sure to give them a minute to cool down. Hardware, particularly cheap hardware, tends to fail when it gets hot, and this can take a while to happen, and, well, it’s been hot.

    Here’s a few things with a high likelihood of failing when you’re away from home:

    • heat, as previously mentioned.
    • running out of disk space. Maybe you’re logging too much, throw some more disk in there and tune down the logging. This can definitely affect SSH, and definitely won’t be fixed by a reboot.
    • OOM failures (or other resource leaks). This isn’t likely to affect your bare metal ssh, but it could. Some things leak memory, and this can lead to cascading process destruction by the OS. In this scenario you’d probably be able to connect to things in the first few minutes after a reboot, though.
    • shitty cabling. Sometimes stuff just falls out of the socket, if it wasn’t plugged in perfectly to begin with. (Heat can also contribute to this one.)
    • reliance on a cloud service that’s currently down. (This can include: you didn’t pay the bill.) Hopefully your OS boot doesn’t fail due to a cloud service, but I’ve definitely seen setups that could.

  • Well people use ansible for a wide variety of things so there’s no straightforward answer. It’s a Python program, it can in theory do anything, and you’ll find people trying to do anything with it. That said, some common ways to replace it include

    • you need terraform or pulumi or something for provisioning infrastructure anyway, so a ton of stuff can be done that way instead of using ansible. Infra tools aren’t really the same thing, but there are definitely a few neat tricks you can do with them that might save you from reaching for ansible.
    • Kubernetes + helm is a big bear to wrestle, but if your company is also a big bear, it’s worth doing. K8s will also solve a lot of the same problems as ansible in a more maintainable way.
    • Containerization of components is great even if you don’t use kubernetes.
    • if you’re working at the VM level instead of the container level, cloud-init can allow you to take your generic multipurpose image and make it configure itself into whatever you need at boot. Teams sometimes use ansible in the cloud-init architecture, but it’s usually doing only a tiny amount of localhost work and no dynamic invetory in that role, so it’s a lot nicer there.
    • maybe just write a Python program or even a shell script? If your team has development skills at all, a simple bespoke tool to solve a specific problem can be way nicer.

  • Really all of these have solutions, but they’re constantly biting you and slowing down development and requiring people to be constantly trained on the gotchas. So it’s not that you can’t make it work, it’s that the cost of keeping it working eats away at all the productive things you can be doing, and that problem accelerates.

    The last bullet is perhaps unfair; any decent system would be a maintainable system, and any unmaintainable system becomes less maintainable the bigger your investment in it. Still, it’s why I urge teams to stop using it as soon as they can, because the problem only gets worse.

  • Sure, I mean, we could talk about

    • dynamic inventory on AWS means the ansible interpreter will end up with three completely separate sets of hostnames for your architecture, not even including the actual DNS name. if you also need dynamic inventory on GCP, that’s three completely different sets of hostnames, i.e. they are derived from different properties of the instances than the AWS names.
    • btw, those names are exposed to the ansible runtime graph via different names i.e. ansible_inventory vs some other thing, based on who even fuckin knows, but sometimes the way you access the name will completely change from one role to the next.
    • ansible-vault’s semantics for when things can be decrypted and when they can’t leads to completely nonsense solutions like a yaml file with normal contents where individual strings are encrypted and base64-encoded inline within the yaml, and others are not. This syntax doesn’t work everywhere. The opaque contents of the encrypted strings can sometimes be treated as traversible yaml and sometimes cannot be.
    • ansible uses the system python interpreter, so if you need it to do anything that uses a different Python interpreter (because that’s where your apps are installed), you have to force it to switch back and forth between interpreters. Also, the python setting in ansible is global to the interpreter meaning you could end up leaking the wrong interpreter into the role that follows the one you were trying to tweak, causing almost invisible problems.
    • ansible output and error reporting is just a goddamn mess. I mean look at this shit. Care to guess which one of those gives you a stream which is parseable as json? Just kidding, none of them do, because ansible always prefixes each line.
    • tags are a joke. do you want to run just part of a playbook? --start-at. But oops, because not every single task in your playbook is idempotent, that will not work, ever, because something was supposed to happen earlier on that didn’t. So if you start at a particular tag, or run only the tasks that have a particular tag, your playbook will fail. Or worse, it will work, but it will work completely differently than in production because of some value that leaked into the role you were skipping into.
    • Last but not least, using ansible in production means your engineers will keep building onto it, making it more and more complex, “just one more task bro”. The bigger it gets, the more fragile it gets, and the more all of these problems rears its head.

  • I can answer this one, but mainly only in reference to the other popular solutions:

    • nginx. Solid, reliable, uncomplicated, but. Reverse proxy semantics have a weird dependency on manually setting up a dns resolver (why??) and you have to restart the instance if your upstream gets replaced.
    • traefik. I am literally a cloud software engineer, I’ve been doing Linux networking since 1994 and I’ve made 3 separate attempts to configure traefik to work according to its promises. It has never worked correctly. Traefik’s main selling point to me is its automatic docker proxying via labels, but this doesn’t even help you if you also have multiple VMs. Basically a non-starter due to poor docs and complexity.
    • caddy. Solid, reliable, uncomplicated. It will do acme cert provisioning out of the box for you if you want (I don’t use that feature because I have a wildcard cert, but it seems nice). Also doesn’t suffer from the problems I’ve listed above.