48.4. Index Locking Considerations

An index access method can choose whether it supports concurrent updates of the index by multiple processes. If the method's pg_am.amconcurrent flag is true, then the core PostgreSQL system obtains AccessShareLock on the index during an index scan, and RowExclusiveLock when updating the index. Since these lock types do not conflict, the access method is responsible for handling any fine-grained locking it may need. An exclusive lock on the index as a whole will be taken only during index creation, destruction, or REINDEX. When amconcurrent is false, PostgreSQL still obtains AccessShareLock during index scans, but it obtains AccessExclusiveLock during any update. This ensures that updaters have sole use of the index. Note that this implicitly assumes that index scans are read-only; an access method that might modify the index during a scan will still have to do its own locking to handle the case of concurrent scans.

Recall that a backend's own locks never conflict; therefore, even a non-concurrent index type must be prepared to handle the case where a backend is inserting or deleting entries in an index that it is itself scanning. (This is of course necessary to support an UPDATE that uses the index to find the rows to be updated.)

Building an index type that supports concurrent updates usually requires extensive and subtle analysis of the required behavior. For the b-tree and hash index types, you can read about the design decisions involved in src/backend/access/nbtree/README and src/backend/access/hash/README.

Aside from the index's own internal consistency requirements, concurrent updates create issues about consistency between the parent table (the heap) and the index. Because PostgreSQL separates accesses and updates of the heap from those of the index, there are windows in which the index may be inconsistent with the heap. We handle this problem with the following rules:

• A new heap entry is made before making its index entries. (Therefore a concurrent index scan is likely to fail to see the heap entry. This is okay because the index reader would be uninterested in an uncommitted row anyway. But see Section 48.5.)

• When a heap entry is to be deleted (by VACUUM), all its index entries must be removed first.

• For concurrent index types, an index scan must maintain a pin on the index page holding the item last returned by amgettuple, and ambulkdelete cannot delete entries from pages that are pinned by other backends. The need for this rule is explained below.

If an index is concurrent then it is possible for an index reader to see an index entry just before it is removed by VACUUM, and then to arrive at the corresponding heap entry after that was removed by VACUUM. (With a nonconcurrent index, this is not possible because of the conflicting index-level locks that will be taken out.) This creates no serious problems if that item number is still unused when the reader reaches it, since an empty item slot will be ignored by heap_fetch(). But what if a third backend has already re-used the item slot for something else? When using an MVCC-compliant snapshot, there is no problem because the new occupant of the slot is certain to be too new to pass the snapshot test. However, with a non-MVCC-compliant snapshot (such as SnapshotNow), it would be possible to accept and return a row that does not in fact match the scan keys. We could defend against this scenario by requiring the scan keys to be rechecked against the heap row in all cases, but that is too expensive. Instead, we use a pin on an index page as a proxy to indicate that the reader may still be "in flight" from the index entry to the matching heap entry. Making ambulkdelete block on such a pin ensures that VACUUM cannot delete the heap entry before the reader is done with it. This solution costs little in run time, and adds blocking overhead only in the rare cases where there actually is a conflict.

This solution requires that index scans be "synchronous": we have to fetch each heap tuple immediately after scanning the corresponding index entry. This is expensive for a number of reasons. An "asynchronous" scan in which we collect many TIDs from the index, and only visit the heap tuples sometime later, requires much less index locking overhead and may allow a more efficient heap access pattern. Per the above analysis, we must use the synchronous approach for non-MVCC-compliant snapshots, but an asynchronous scan is workable for a query using an MVCC snapshot.

In an amgetmulti index scan, the access method need not guarantee to keep an index pin on any of the returned tuples. (It would be impractical to pin more than the last one anyway.) Therefore it is only safe to use such scans with MVCC-compliant snapshots.