Andal.LND/edx-platform
6
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Merge pull request #4899 from edx/dhm/mongo_indices

Browse Source 
Refactored expected indices after email w/ mongo
This commit is contained in:
Don Mitchell
2014-08-25 10:19:27 -04:00
parent b3f4ce2312 ceae081cc7
commit e4aaf4975c
1 changed files with 9 additions and 53 deletions
Download Patch File Download Diff File Expand all files Collapse all files

62
mongo_indexes.md
View File

@@ -20,16 +20,14 @@ which can be `uploadDate`, `display_name`,
Replace existing index which leaves out `run` with this one:
```
ensureIndex({'_id.tag': 1, '_id.org': 1, '_id.course': 1, '_id.category': 1, '_id.run': 1})
ensureIndex({'content_son.tag': 1, 'content_son.org': 1, 'content_son.course': 1, 'content_son.category': 1, 'content_son.run': 1})
ensureIndex({'_id.org': 1, '_id.course': 1, '_id.name': 1}, {'sparse': true})
ensureIndex({'content_son.org': 1, 'content_son.course': 1, 'content_son.name': 1}, {'sparse': true})
ensureIndex({'_id.org': 1, '_id.course': 1, 'uploadDate': 1}, {'sparse': true})
ensureIndex({'_id.org': 1, '_id.course': 1, 'display_name': 1}, {'sparse': true})
ensureIndex({'content_son.org': 1, 'content_son.course': 1, 'uploadDate': 1}, {'sparse': true})
ensureIndex({'content_son.org': 1, 'content_son.course': 1, 'display_name': 1}, {'sparse': true})
```
Note: I'm not advocating adding one which leaves out `category` for now because that would only be
used for `delete_all_course_assets` which in the future should not actually delete the assets except
when doing garbage collection.
Remove index on `displayname`
modulestore:
============
@@ -41,7 +39,7 @@ and no field is omitted.
Because we often query for some subset of the id, we define this index:
```
ensureIndex({'_id.tag': 1, '_id.org': 1, '_id.course': 1, '_id.category': 1, '_id.name': 1, '_id.revision': 1})
ensureIndex({'_id.org': 1, '_id.course': 1, '_id.category': 1, '_id.name': 1})
```
Because we often scan for all category='course' regardless of the value of the other fields:
@@ -51,54 +49,12 @@ ensureIndex({'_id.category': 1})
Because lms calls get_parent_locations frequently (for path generation):
```
ensureIndex({'_id.tag': 1, '_id.org': 1, '_id.course': 1, 'definition.children': 1})
```
Remove these indices if they exist as I can find no use for them:
```
{ "_id.course": 1, "_id.org": 1, "_id.revision": 1, "definition.children": 1 }
{ "definition.children": 1 }
```
NOTE, that index will only aid queries which provide the keys in exactly that form and order. The query can
omit later fields of the query but not earlier. Thus ```modulestore.find({'_id.org': 'myu'})``` will not use
the index as it omits the tag. As soon as mongo comes across an index field omitted from the query, it stops
considering the index. On the other hand, ```modulestore.find({'_id.tag': 'i4x', '_id.org': 'myu', '_id.category': 'problem'})```
will use the index to get the records matching the tag and org and then will scan all of them
for matches to the category.
To find out if any records have the wrong id structure, run
```
db.fs.files.find({uploadDate: {$gt: startDate, $lt: endDate},
$where: function() {
var keys = ['category', 'name', 'course', 'tag', 'org', 'revision'];
for (var key in this._id) {
if (key != keys.shift()) {
return true;
}
}
return false;
}},
{_id: 1})
ensureIndex({'definition.children': 1}, {'sparse': true})
```
modulestore.active_versions
===========================
```
ensureIndex({'org': 1, 'offering': 1})
ensureIndex({'org': 1, 'course': 1, 'run': 1}, {'unique': true})
```
modulestore.structures
======================
```
ensureIndex({'previous_version': 1})
```
modulestore.definitions
=======================
```
ensureIndex({'category': 1})
```