Skip to content

Commit

Permalink
Merge branch 'main' into update-hubspot-link
Browse files Browse the repository at this point in the history
  • Loading branch information
hengm3467 authored Dec 17, 2024
2 parents 5e45a78 + 85a176b commit 3f25f54
Showing 1 changed file with 0 additions and 32 deletions.
32 changes: 0 additions & 32 deletions operate/meta-backup.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -147,35 +147,3 @@ Parameters to `risectl meta restore-meta` should be:
* `--hummock-storage-url s3://state_bucket`. Note that the `hummock+` prefix is stripped.
* `--hummock-storage-directory state_data`.
4. Configure meta service to use the new meta store.

## Access historical data backed up by meta snapshot

Meta snapshot is used to support historical data access, also known as time travel query.

Use the following steps to perform a time travel query.

1. List all valid historical point-in-time (i.e., epoch) for a table. For example to query the table of id 6:
```sql
SELECT state_table_info->'6'->>'safeEpoch' as safe_epoch,state_table_info->'6'->>'committedEpoch' committed_epoch from rw_meta_snapshot;
```
Example output:
```bash
safe_epoch | committed_epoch
-----------------+------------------
7039353459507200 | 7039354678542336
7039354678542346 | 7039622397886464
```
Choose an epoch to query. Valid epochs are within range \[`safe_epoch`,`committed_epoch`\], e.g. \[7039353459507200, 7039354678542336\] or \[7039354678542346, 7039622397886464\].
2. Set session config `QUERY_EPOCH`. By default, it's 0, which means disabling historical query.
```sql
SET QUERY_EPOCH=[chosen epoch];
```
Then, batch queries in this session return data as of this epoch instead of the latest one.
3. Disable historical query.
```sql
SET QUERY_EPOCH=0;
```

<Note>
RisingWave only supports historical data access at a specific point in time backed up by at least one meta snapshot.
</Note>

0 comments on commit 3f25f54

Please sign in to comment.