• 0 Posts
  • 132 Comments
Joined 1 year ago
cake
Cake day: July 1st, 2023

help-circle


  • It’s good practice to run the deployment pipeline on a different server from the application host(s) so that the deployment instances can be kept private, unlike the public app hosts, and therefore can be better protected from external bad actors. It is also good practice because this separation of concerns means the deployment pipeline survives even if the app servers need to be torn down and reprovisioned.

    Of course you will need some kind of agent running on the app servers to be able to receive the files, but that might be as simple as an SSH session for file transfer.



  • That’s probably okay! =) There’s some level of pragmatism, depending on the sort of project you’re working on.

    If it’s a big app with lots of users, you should use automation because it helps reliability.

    If there are lots of developers, you should use automation because it helps keep everyone organised and avoids human mistakes.

    But if it’s a small thing with a few devs, or especially a personal project, it might be easier to do without :)


  • Sure, but having a hands-off pipeline for it which runs automatically is where the value is at.

    Means that there’s predictability and control in what is being done, and once the pipeline is built it’s as easy as a single button press to release.

    How many times when doing it manually have you been like “Oh shit, I just FTPd the WRONG STUFF up to production!” - I know I have. Or even worse you do that and don’t notice you did it.

    Automation takes a lot of the risk out.


  • I’m sure there’s nothing wrong with the program at all =)

    Modern webapp deployment approach is typically to have an automated continuous build and deployment pipeline triggered from source control, which deploys into a staging environment for testing, and then promotes the same precise tested artifacts to production. Probably all in the cloud too.

    Compared to that, manually FTPing the files up to the server seems ridiculously antiquated, to the extent that newbies in the biz can’t even believe we ever did it that way. But it’s genuinely what we were all doing not so long ago.







  • Technical requirements are often ambiguous when written as free text, the way someone would speak them, because as you have discovered the free text fails to capture where the linguistic stress would be that disambiguates in speech.

    Instead, I suggest using a format that is more suited to text.

    I would recommend a table. Email the customer back with your current interpretation of the requirements, with a column for outcome and a column for value. Ask them to check and sign off on the table, or to correct the table where it is wrong.

    Example:

    Outcome Value
    NULL x
    Complete x
    Cancelled x
    (Other) x

    There are edge-cases with if outcome can be "Complete or Cncelled