MongoDB聚合计数异常：总数与分项计数不符问题排查

Ah, I see the issue here — those missing 5 events are falling through the cracks because of how your initial $match stages interact with array fields in MongoDB. Let's break this down and fix it.

Why the Discrepancy Happens #

Your three aggregation commands have a key flaw in the first $match step:

• Command 1 correctly counts all events with an _id (100 total) because it doesn't filter based on field existence at the document level.
• Command 2 uses {"events.field": {$exists: true}} as the first match: this returns any document that has at least one event with a field. When you $unwind and filter again, you count only those events with field (80), but these documents may also contain events without field.
• Command 3 uses {"events.field": {$exists: false}} as the first match: this only returns documents where every single event in the array lacks a field (or the events array is empty/missing). It completely ignores documents that have a mix of events with and without field.

The missing 5 events are exactly those: events without field that live in documents that also have events with field — they get excluded from both Command 2 and Command 3.

How to Recover the Missing 5 Events #

Option 1: Get Full, Accurate Counts in One Aggregation #

Instead of splitting the count into three commands, run a single aggregation that counts all events by field existence without filtering out documents early:

db.collection.aggregate([
  // Match only documents with valid events (matches Command 1's logic)
  {$match: {"events._id": {$exists: true}}},
  {$unwind: "$events"},
  // Ensure we only count events with an _id (same as Command 1)
  {$match: {"events._id": {$exists: true}}},
  // Group by whether the event has a field or not
  {$group: {
    _id: {$cond: [{$exists: ["$events.field", true]}, "has_field", "no_field"]},
    count: {$sum: 1}
  }}
])

This will return two groups: has_field (80) and no_field (20, which is 15 + 5). The 20 total for no_field includes both the events from all-no-field documents and the missing 5 from mixed documents.

Option 2: Directly Query the Missing Events #

If you want to inspect or count just the missing 5 events, use this aggregation to target them specifically:

db.collection.aggregate([
  // Find documents that have BOTH:
  // - At least one event with a field, AND
  // - At least one event without a field (but with an _id)
  {$match: {
    "events._id": {$exists: true},
    "events.field": {$exists: true},
    "events": {$elemMatch: {"field": {$exists: false}, "_id": {$exists: true}}}
  }},
  {$unwind: "$events"},
  // Filter to just the events without a field (but with _id)
  {$match: {"events.field": {$exists: false}, "events._id": {$exists: true}}},
  // Count them (or remove this stage to see the actual events)
  {$group: {_id: null, missing_event_count: {$sum: 1}}}
])

This will directly return the count of 5 missing events, or show you the actual event documents if you omit the final $group stage.

Key Takeaway #

When working with array fields in MongoDB, document-level $match conditions like {"array.field": {$exists: ...}} don't filter individual array elements — they filter entire documents based on whether any (or all, for $exists: false) elements meet the condition. Always $unwind first (or use array operators like $elemMatch) if you need to target individual array elements accurately.

内容的提问来源于stack exchange，提问作者Panikos Stavrou