video1.0<iframe src="https://www.loom.com/embed/48bb136113c14fa881559b259e4732c4" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/os/loom-media-content/loom/95cdfc1b-36fc-40fe-8132-df3029db3286/0a456184-9cb5-4e0b-bc1c-78998f7bf2c4/sessions-thumbnails-48bb136113c14fa881559b259e4732c4-48db6805be805b72.gif/binary?authToken=eyJraWQiOiJtaWNyb3MtZ3JvdXAvdGRwLW9zLWFyY2hldHlwZS9wN2syNWplYmdrZ2R0YnUxIiwiYWxnIjoiUlMyNTYifQ.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.ZOyokpUYAGMC3dLQJb2zgsEO8nLV9GSvISE73BSDbwaK2uj3G18IdEg6l3oSHL0HAlG8QmcbxnJoi5h9av9batC5NqNxCM2ufHyzyEklaZpyuiZG0PdpDn0Xrd8UQSj-N-ZPKZTkQGi0UZ-Pb_QpxAGyQ2H5_KJJF7B9_JG_GZY2fzehCX9gVDgNf9ZzDioYJT6-q3zITFUsIi8IVuGR5Ihv1UkQF1H8emIIH5OjpEj6rKQFeXBjMn0gQrbUuuWPBfnAbzWKgRSyOghLn-d6BWf21tlU7t6NYWKyYT4DuVVtQjGt_t6cIG6pABLHHAy2hnaR5n01qdAmdO5pIGrN_w&Expires=1792530834&Signature=ZcRWKBh1REk2z2O6N%7EpMs20b0SMolIRLRJdbFLz1T8bqwIS%7ElyKitrS-qBgur8nHY9TWbfRAHdPmAiBb4GytCrrxT-cyXc0thMpYb5YqvXE0vi2K2B3Trb8P7TEwjvx3qk88NlTb1qnLYyDm7dZhWvZzTqWMEwMTsfXloA%7EwL7zKMWA7TCJN%7EvI2Lfyb7i6y3rfWZjOOYCIlEN27a7gaZGeANEM4HE4zsZiOvxlglaAWE1XqOHPXgEgSObriz%7E8J9GBY7irVaqYvqlLNSgaY6KsD7VRpj563Nuee1NN04Rg-bQy03fn-DRbgNnxpfWBkMYV8ArXGDzXuoXj3k68jcg__&Key-Pair-Id=KQOSYIR44AIC0320.22This Loom explains how to fix slow production time series queries using a fork stroke fix stroke delete approach to avoid breaking tables and locking. The creator uses Copilot agent mode to fork the production TigerCloud timescale database, creating db_63453_test as an independent copy in US Virginia, then reproduces the issue on the fork and compares explain-analyzed plans before and after. The sequential scan baseline was 0.748 ms with 66 shared buffer hits, while the corrected indexed plan became 0.155 ms with 3 shared buffer hits, improving execution time about 4.8x. After confirming no fatal logs, the fork is deleted and monitoring shows clean checkpoints, with expected fork shutdown errors.