No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.
No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.
No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.
No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.
No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.
No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.
No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.
No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.
No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.
No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.
No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.
No spike in CPU. Error budgets looked fine at a glance. Users still reported blank pages in a thin slice of traffic. Logs only showed upstream resets with no clear application exception. The culprit was a stale keep-alive timeout between nginx and the app. Aligning idle timeouts stopped the intermittent 502s within an hour. We also added a dashboard for upstream reset reasons so the next page is faster.