diff --git a/.claude/commands/migrate-ccdb.md b/.claude/commands/migrate-ccdb.md index 15670bd84fd..19ea03bd26d 100644 --- a/.claude/commands/migrate-ccdb.md +++ b/.claude/commands/migrate-ccdb.md @@ -195,10 +195,10 @@ pre-existing behaviour — every distinct timestamp may yield a different object Pick it from the object's real validity, and only then: -| Object changes ... | Uniformity | Declare with | -| --- | --- | --- | -| within a run (calibrations, drift velocity) | timestamp (default) | `DECLARE_SOA_TIMESTAMPED_TABLE` | -| per run or per period (geometry, material, per-period calibrations) | `aod::BCs` / `aod::bc::RunNumber` | `DECLARE_SOA_UNIFORM_TABLE` | +| Object changes ... | Uniformity | Declare with | +|---------------------------------------------------------------------|-----------------------------------|---------------------------------| +| within a run (calibrations, drift velocity) | timestamp (default) | `DECLARE_SOA_TIMESTAMPED_TABLE` | +| per run or per period (geometry, material, per-period calibrations) | `aod::BCs` / `aod::bc::RunNumber` | `DECLARE_SOA_UNIFORM_TABLE` | Worked examples in the tree: `aod::TpcCalibCCDBObjects` keeps the timestamp default because the TPC drift velocity genuinely varies within a run; `aod::GeomCCDBObjects` and @@ -297,12 +297,12 @@ The fetcher downloads once into a shm cache and the column stores `(handle, segm Measured on `propagationServiceV2` (`HF_LHC23_pass4_Thin_small_2P3PDstar`, `daily-20260922-0000-1`), migrated wagon 61527 against un-migrated baseline wagon 61526: -| | baseline 766059 | migrated 766456 | -| --- | --- | --- | -| propagation device `cpuUsedAbsolute` | 3,972,078 | 2,866,181 (**−28 %**) | -| propagation device peak PSS | 160 MB | 168 MB | -| train peak PSS | 1340 MB | 1366 MB | -| CCDB queries from the task device | 3.03 MB fetched | 0 — all via the fetcher | +| | baseline 766059 | migrated 766456 | +|--------------------------------------|-----------------|-------------------------| +| propagation device `cpuUsedAbsolute` | 3,972,078 | 2,866,181 (**−28 %**) | +| propagation device peak PSS | 160 MB | 168 MB | +| train peak PSS | 1340 MB | 1366 MB | +| CCDB queries from the task device | 3.03 MB fetched | 0 — all via the fetcher | Quote the CPU number when justifying a migration; do not promise memory.