Weird deployment procedures


  • kills Dumbledore

    Share your stories of hoops jumped through, strange request formats or whatever. I'll start off with a minor one

    Here, to get an update to a site pushed to UAT or live we have to email the helpdesk with a link to an instructions .txt document on a network share. This documents where the new files are (another network share), where they should go and any other changes like SQL to be run or web.config changes. The infrastructure team follows the instructions and don't ask questions. They're not devops by any stretch.

    The old files are never removed, so I can look at instructions for a deployment from 2012 and also follow the gradually changing formats of the naming system. The folder for the files to deploy is also never cleared, so there are a lot of old, obsolete versions of binaries in folders with names like "ourWebSite 2.03". It's like a really shitty, binary only ZFSC


  • I survived the hour long Uno hand

    @Jaloopa We used to have to svn diff, export any changed files, put that on a network share with a very specific name scheme (so an automated script could find it), put any SQL changes in a similarly specific named folder, and then print out an excel spreadsheet CR form and get it signed off manually. Then we'd have the release coordinator grab all the files and copy them to the staging server to verify, then copy them into prod.

    Before that we didn't have source control, but I never experienced that system, thank god.


  • kills Dumbledore

    @Yamikuronue said in Weird deployment procedures:

    automated script

    That would improve a lot of things around here


  • I survived the hour long Uno hand

    @Jaloopa Yeah, someone threw together a batch script they could run after they RDP into the servers to help keep his work life sane.


Log in to reply