Drupal Migration — Performance Monitor

Bottom line — Drupal 10 closed out its ~4-week run as the most reliable phase to date (664ms avg response, 0.78 apdex, 0.36% error rate — the lowest error rate of the three mature phases), and today's Drupal 11 upgrade completed cleanly with a 26-minute maintenance window (05:58–06:26 Chicago) and no lingering issues visible in the post-deploy traffic. Drupal 11 itself is far too new to judge: the phase window has been open for well under a minute at data-pull time, so its 9-request, 1195ms, 0.61-apdex read is noise, not signal, and should not be compared against D10/D8 yet.

• D11 early read (thin sample, not yet meaningful): only 9 requests logged since the 07:00 Chicago cutoff — several orders of magnitude below D10's 3.11M-request baseline. The elevated resp (1195ms) and low apdex (0.61) are almost certainly single-request/cold-cache artifacts, not a real performance signal. Re-check tomorrow once volume builds.
• Maintenance window excluded from aggregates but visible in the time series: D10's phase cutoff (05:00 Chicago) and D11's phase start (07:00 Chicago) deliberately exclude the 05:00–07:00 deploy/maintenance window from both phase aggregates. That window still shows up as a traffic dip and error spike in the 24h series — the 05:59 hourly bucket shows a 6.23% error rate (vs a sub-1% baseline) and reduced request volume, consistent with the maintenance cutover rather than an ongoing production issue.
• D10 was the strongest mature phase on reliability: error rate improved monotonically D8 (0.39%) → D9 (0.42%, slightly worse) → D10 (0.36%, best of the three) while response time stayed essentially flat (679 → 699 → 664ms) and apdex held steady (0.79 → 0.77 → 0.78). No regression carried from D9 into D10.
• 7-day trend is normal weekly cyclicality, not an anomaly: daily request volume on the weekend (Jul 25–26, ~35K/day) is roughly a quarter of the weekday volume (Jul 27–31, ~110K–160K/day), which is the expected pattern. Response time drifted up modestly across the week (515ms Jul 25 → 671ms Jul 31) but stayed within the phase's normal range.
• 24h view: aside from the maintenance-window dip late in the window, hourly response times and error rates tracked close to the D10 phase baseline for the rest of the day, with no other notable spikes.

Method & caveats

• Cutovers. Drupal 9 / PHP 8 went live May 23, 2026; Drupal 10 Jul 4, 2026 — marked on every chart that spans them.
• Drupal 10 is an early read (~5 days). Its phase card uses the Jul 4–9 window directly.
• Recent spike. Jul 8 shows an isolated response jump (≈2× the prior day), coinciding with elevated reports/progress-API latency that day — an operational blip, not a migration regression.
• Source & retention. New Relic timeslice metrics (HttpDispatcher, Apdex, Errors/all), ~13-month retention — the full arc back to mid-2025 is covered. Percentiles (p95/p99) live in 8-day event data and are reported in the point-in-time reviews, not here.
• Refresh. Regenerated daily by a scheduled routine that re-queries New Relic and redeploys to this same URL.