At a glance
Top 10 Slow Operations is a ranked table of the ten slowest database operations recorded over the last 24 hours, broken down row by row. It is the detail behind the Slow Ops (15m, >100ms) count: where that card tells you how many operations were slow, this one tells you exactly which ones, so you know what to index, rewrite, or investigate first. Each row identifies a slow operation by its query shape, the collection it touched, and its execution time.
What it tracks
The table lists the Top 10 Slow Operations over a rolling 24-hour window, broken down by row. Each entry is drawn from the database profiler’ssystem.profile collection, ranked by execution time (millis), so the slowest operation sits at the top. A typical row carries the query shape (the namespace and predicate), the collection, the duration, and the plan summary, which is where a COLLSCAN (full collection scan) or a missing index reveals itself. Read this card immediately after a slow-ops alert: if one shape appears repeatedly near the top, that single query is almost always the fix, usually a new index or a rewrite. The 24-hour window is deliberately wider than the 15-minute alert window so that intermittent slow operations (a nightly batch job, a periodic report) are still captured rather than scrolling out of view.
Reconciling against the source
To reproduce this list natively, rundb.system.profile.find().sort({ millis: -1 }).limit(10) in mongosh against a profiled database, or use the slow-query lines in the mongod log. On MongoDB Atlas, the Query Profiler and Performance Advisor present the same ranked slow-query view grouped by shape. Confirm the profiling level and slowms with db.getProfilingStatus() so the native list and this card use the same threshold.