Publicly exposed server setup … at vendor's request?



  • Continuing the discussion from A prime example of why I tend to avoid *nix:

    @abarker said:

    [The server] was spun up at the request of the last vendor that redesigned our corporate websites.

    A few years ago my employer had the corporate websites completely redesigned by a outside vendor. One of the stipulations was that the new sites should be easily maintainable. That is, if a phone number or image or text needed to be updated, we shouldn't need to contact the vendor. Another consideration is that we were at the time a 100% Microsoft company – aside from the server that allowed users to get email on their Blackberries.

    With those considerations in mind, the executives selected a vendor that specialized in WordPress sites. Strangely enough, they selected a vendor that specialized in *nix instances of WordPress, and had never touched a Windows instance of WordPress. Shouldn't be a problem, they thought. And at first, it appeared that they were right.

    But then came the issues. Some pages weren't loading, unless you were connected to the network. Some images were never updating, no matter what you did. The vendor tried a few things, but finally threw his hands in the air and said, "I can't make it work on Windows. I'm a *nix guy. You need a *nix box"

    And with that, he was given the power to spin up a new server on our network. He was allowed to completely configure a new VM as he saw fit. The only oversight he received was from a team of Windows guys who had little, if any, *nix experience.

    So was born our one, solitary *nix box. Since the Blackberry email server was been forcefully retired a few years back, this server now stands as the lone stalwart against the tide of Microsoft domination in our corporate network.



  • @abarker said:

    "I can't make it work on Windows. I'm a *nix guy. You need a *nix box another vendor"

    FTFH



  • Because one like isn't enough. We've all had a good chuckle in the office here, thanks for that.



  • @thegoryone said:

    Also, by any chance was the Windows server running -shudder- PHP for IIS? That would explain the idiocy. Vendor used to Apache. Would explain a lot, actually.

    Maybe? I was just starting at the company at the time, so I was focused on other things.



  • I ran WordPress on Windows for awhile. As I recall:

    1. The core code was pretty solid, but virtually every plug-in had problems with Windows (usually wrong assumptions about paths)

    2. The original WordPress installer (waaay back when I first set it up) marked the MySQL database as being a non-UTF-8 character set, then stored UTF-8 in it anyway. Meaning, when I did the DB import to the Windows server, the characters got all fucked and had to be manually fixed-up. IIRC this was actually a bug in WordPress when it originally creates the tables it uses, the move to Windows just exposed the bug.

    3. The Windows hosting I had was some shitty hosting company that did a shitty job and were fuck-ups and goddamned fuck them right up the ass.


  • Java Dev

    @blakeyrat said:

    (usually wrong assumptions about paths)

    How'd they manage that, don't slashes actually work on windows? Or is that only apache and not IIS?



  • It's been years, I don't remember the details.

    My point was, the core WordPress developers actually do testing in Windows; virtually none of the plug-in authors do. Whether that's a problem for you personally running WordPress really depends on how many plug-ins you use.



  • They do. But paths are not separated by colons, drive letters are a thing that can possibly exist, you can't (by default) have filenames that differ only by case or have <>|\/*?": or are the name of a legacy DOS device, WordPress is not installed directly in /var/www, ...

    Oh, and there is no ".htaccess", only "the Metabase" or its successor "Web.config".


  • Java Dev

    Sounds like a list of attributes I'd avoid on linux as well...



  • Is this a reported thing?


  • FoxDev

    Yes, and it's been around for ages


  • kills Dumbledore

    Posts aren't rebaked when an avatar changes, so the mini avatar doesn't change.

    Doesn't explain why there isn't just a /users/blakeyrat/currentavatar.png or something...



  • @Jaloopa said:

    Doesn't explain why there isn't just a /users/blakeyrat/currentavatar.png or something...

    The fact that we are are on Duckhorse should explain that. And, even if there was, there would likely be client side caching issues.



  • There is. The last segment of the URL is an upload ID. If you specify one that was never used by that user, like 1.png, then it automatically redirects you to the latest version instead. Of course, this creates a new roundtrip, with a new server request, and we know how much Discourse loves server requests...


  • Discourse touched me in a no-no place

    @TwelveBaud said:

    /#{user.username_lower}

    Now we know where Project Spartan got the insane idea to display the URL in all lowercase letters!



  • Also, I just saw this post about why the avatars aren't updating at all anymore, and died a little inside.



  • Makes me wonder how many avatar changes it'll take before you need a 1TB database only for them.


Log in to reply