No. 0201 min read
Your installer finished. Is the damned system working?
At Raymond James, Splunk and Dynatrace helped cut critical-error diagnosis and remediation time by 40%.
No. 0072 min read
At sixteen I moved Advertisnet's services from Windows NT to Linux/BSDi. The servers were only part of it, folks. Somebody had to answer the damned phone.
At sixteen I was still in high school, and I migrated Advertisnet's entire service stack from Windows NT[1]Ref. 1Windows NT en.wikipedia.org to Linux[2]Ref. 2Linux kernel.org/BSDi[3]Ref. 3BSDi en.wikipedia.org, built redundant user databases for unified authentication, shipped a Perl[4]Ref. 4Perl perl.org CGI[5]Ref. 5CGI en.wikipedia.org support control panel, and grew the support team from zero.
I don't imagine my age reassured anyone whose Internet service I was moving. My age wasn't what made the damned thing run.
The operating systems get the attention. Meanwhile, somebody has to help the customer. You can replace every server and leave everybody with the same problems, plus some exciting new ones. I'd still want to know whether anybody could use the machine after it booted.
I shipped the support control panel in Perl CGI. The Common Gateway Interface lets a web server hand a request to a program and return its output to the browser. Perl was good at that glue work. It could take the input and inspect or change system state, then give the browser something useful back. I liked it for that.
Growing the support team from zero belonged to the same job. Once other people need to work on a system, the stuff one person happens to remember is no longer good enough.
Giving the next person a login doesn't give them everything that's in your head. Probably grateful for that, in my case.
They need to distinguish an account problem from a service problem, and hand off what they've already checked. Otherwise the customer repeats the story, the next technician repeats the tests, and everybody calls it thoroughness.
I'd test that handoff as seriously as the software: can somebody else find the account, understand its state, and take the next safe action without calling me?
Otherwise I'm still the undocumented part of the system that everybody has to call, even though we've removed the old operating system. I'd want that sorted out before calling the fucking migration finished.
No. 0201 min read
At Raymond James, Splunk and Dynatrace helped cut critical-error diagnosis and remediation time by 40%.
No. 0091 min read
Fifty servers, fourteen years, 99.999% availability.
No. 0543 min read
A timestamp, a cron line, an exit code, an invisible Unicode character.
No. 0531 min read
A shell that can undo the last thing you did.