
Half the Time Went to Logs Nobody Read
I timed OrphanedSid at about a second per object on my lab DC. Half went to Debug log lines it formatted and threw away, and much of the rest to turning SIDs into names and back.

I timed OrphanedSid at about a second per object on my lab DC. Half went to Debug log lines it formatted and threw away, and much of the rest to turning SIDs into names and back.

In 2025 I wrote a whole Pester post about Test-PathTraversal. No OrphanedSid command ever called it, it had two bugs, and 1.1.0 deleted it with 42 other functions nothing reached.

I built OrphanedSid’s log for audits, then reran 1.0.1 on my lab DC. A change number missed 10 of 36 lines, a config file split one run across two files, and a reused log kept one night.

I piped one OU into Find-OrphanedSid, the way the README showed. OrphanedSid 1.0.x dropped the OU and searched the whole domain, and in my lab it still came back with the right two orphans.

I gave a lab group two orphaned SIDs and removed both. OrphanedSid 1.0.x backed the group up once per SID, so restore brought back only one, and a long DN on Windows PowerShell 5.1 got no backup at all.

The night 1.0.1 shipped, I filed a bug: OrphanedSid called a SID orphaned whenever its lookups failed, so a timed-out DC could put live access on the removal list. 1.1.0 added a third answer, Unverified.

OrphanedSid’s first release failed before it reached the PowerShell Gallery, and the retry went up fine. Fifteen minutes later I found it couldn’t restore a single object, and 1.0.1 came within the hour.

OrphanedSid’s guides went through four rewrites in five days, and every pass after the first found code that didn’t do what the docs said. Checking the docs was one of the better tests I ran on the module.

OrphanedSid went to the PowerShell Gallery with 2,400 passing tests, and a lot of them couldn’t fail. Rewriting the suite cut it to 1,374 tests, took coverage from 62% to 94%, and turned up twenty issues.

OrphanedSid passed every test locally and on a lab domain controller, then failed every job on its first GitHub Actions run. Four runs later it was green, and almost none of the fixes touched the module.