Github user JamesRTaylor commented on a diff in the pull request:

    https://github.com/apache/phoenix/pull/12#discussion_r17406778
  
    --- Diff: 
phoenix-core/src/main/java/org/apache/phoenix/schema/MetaDataClient.java ---
    @@ -462,6 +477,71 @@ public MutationState createTable(CreateTableStatement 
statement, byte[][] splits
             return connection.getQueryServices().updateData(plan);
         }
     
    +    public long updateStatistics(UpdateStatisticsStatement 
updateStatisticsStmt) throws SQLException {
    +        String tableName = 
updateStatisticsStmt.getTable().getName().getTableName();
    +        // Check before updating the stats if we have reached the 
configured time to reupdate the stats once again
    +        long minTimeForStatsUpdate = 
connection.getQueryServices().getProps()
    +                .getLong(StatisticsConstants.MIN_STATS_FREQ_UPDATION, 
StatisticsConstants.DEFAULT_STATS_FREQ_UPDATION);
    +        // TODO : Check if we need the table key type of table name here. 
    +        // May be we can avoid multiple calls from the 
    +        // same connection
    +        byte[] tenantIdBytes = QueryConstants.EMPTY_BYTE_ARRAY;
    +        // TODO : If tenantId is not null we may have to get the actual 
table name (PTable.getPhysicalName)
    +        if (connection.getTenantId() != null) {
    +            tenantIdBytes = connection.getTenantId().getBytes();
    +        }
    +        byte[] schemaNameBytes = QueryConstants.EMPTY_BYTE_ARRAY;
    +        if (connection.getSchema() != null) {
    +            schemaNameBytes = Bytes.toBytes(connection.getSchema());
    +        }
    +        try {
    +            if (inProgress.get() > 0) {
    +                // Already in progress
    +                return 0;
    +            }
    +            inProgress.incrementAndGet();
    +            // Always invalidate the cache
    +            
connection.getQueryServices().clearCacheForTable(tenantIdBytes, 
schemaNameBytes, Bytes.toBytes(tableName));
    +            String schema = Bytes.toString(schemaNameBytes);
    +            // Clear the cache also. So that for cases like major 
compaction also we would be able to use the stats
    +            updateCache(schema, tableName, true);
    +            String query = "SELECT " + "LAST_STATS_UPDATE_TIME_IN_MS " + " 
FROM " + SYSTEM_CATALOG_SCHEMA + "."
    +                    + SYSTEM_STATS_TABLE + " WHERE "
    +                    + TABLE_NAME + "='" + tableName + "' AND " + 
COLUMN_NAME + " IS NULL AND " + REGION_NAME
    +                    + " IS NULL";
    +            ResultSet rs = 
connection.createStatement().executeQuery(query);
    +            long lastUpdatedTime = 0;
    +            if (rs.next()) {
    +                lastUpdatedTime = rs.getLong(1);
    +            }
    +            long currentTime = TimeKeeper.SYSTEM.getCurrentTime();
    --- End diff --
    
    Let the server do the update of the last_stats_updated (as it needs to do 
it anyway). Just make sure to call your invalidate_table call after the analyze 
call (in ConnectionQueryServicesImpl) so that the PTable on the server gets 
it's timestamp updated.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---

Reply via email to