# Master Aggregator Redundancy Error Codes During Master Aggregator (MA) failover, the voting members elect a new MA. The election protocol is internal; applications do not receive a SQL error code for each vote. Instead, a client can receive a connection error or an MA-role error while the old MA is shutting down or the new MA is being elected. Configure the application to retry transient failures through the load-balanced MA endpoint. For a query that cannot be forwarded or retried automatically, drop the connection, reconnect, and retry the operation according to the application's retry and transaction-safety requirements. ## Client and Failover Errors The following errors can occur during MA failover. | Error code | Available in DB release 8.9/ 9.0 / 9.1 | When it occurs | Recommended action | | --------------------------------------- | -------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------- | | ER\_SERVER\_SHUTDOWN | 1053 | The previous MA is shutting down. This can also be returned when a DDL term check runs during shutdown. | Retry through the load-balanced MA endpoint. | | ER\_CONNECT\_TO\_MASTER | 1218 | A child aggregator cannot connect to the current MA, typically while a new MA is being elected. The message is « Error connecting to master: %s ». | Retry the request after the new MA is available. | | ER\_DISTRIBUTED\_NOT\_MASTER | 1752 | The request reached a node that is not the current MA, including after another voting member was elected. | Retry on the current MA through the load-balanced endpoint | | ER\_DISTRIBUTED\_NOT\_MASTER\_ANYMORE | 2076 | The node that received the request is no longer the MA, for example, after a leadership transfer. | Retry on the new MA. | | ER\_DISTRIBUTED\_MASTER\_TERM\_REJECTED | 2787, first appearing in 9.0.2 | A DDL operation was rejected because an MA failover occurred; the request's DDL term is behind the node's consensus term. | Retry the DDL operation on the new MA. | A driver can also report a lost connection, such as « Lost connection to server during query». This should be handled as a transient failover error: reconnect before retrying the operation. ## Consensus Membership and Leadership Errors These errors apply to administrative operations that manage consensus membership or leadership. They are not SQL errors returned for individual election votes. The consensus-family numeric codes differ between the releases 8.9 and 9.0/9.1. The family beginning with `ER_CONSENSUS_BROKEN_QUORUM` is absent in 8.9.1 and appears from 8.9.19. | Error code | 8.9.1 | 8.9.19–8.9.54 | 9.0 / 9.1 | When it occurs | Recommended action | | ---------------------------------------------- | ------------- | ------------- | --------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------- | | ER\_NODE\_IS\_VOTING\_MEMBER | 2481 | 2481 | 2481 | Consensus is being disabled while the node is still a voting member. | Remove the node from voting membership or re-add it as a follower before setting consensus\_enabled to OFF. | | ER\_CONSENSUS\_NOT\_ENABLED | 2482 | 2482 | 2482 | An operation requiring consensus is attempted while consensus is disabled. This can also occur when consensus is enabled while a reference database is not sync durable. The client message is then the custom durability text, not only the generic consensus-disabled text. | Enable consensus after satisfying its prerequisites; for a durability-related message, make the reference database sync durable. | | ER\_CONSENSUS\_BROKEN\_QUORUM | Not available | 2749 | 2748 | Detach, demote, or remove would leave no majority of active voting members. | Do not perform the operation until it preserves quorum. | | ER\_CONSENSUS\_NON\_VOTING\_MEMBER | Not available | 2750 | 2749 | A voting-member operation is attempted on a non-voting member, such as a follower CA. | Run the operation on a voting member or add the node to voting membership first. | | ER\_CONSENSUS\_DETACHED\_VOTING\_MEMBER | Not available | 2751 | 2750 | A voting member is detached when it is already detached. | Check the member state before retrying the detach operation. | | ER\_CONSENSUS\_ATTACHED\_VOTING\_MEMBER | Not available | 2752 | 2751 | A voting member is attached when it is not detached. | Check the member state before retrying the attach operation. | | ER\_CONSENSUS\_ATTACH\_VOTING\_MEMBER\_TIMEOUT | Not available | 2753 | 2752 | Warning: ATTACH VOTING MEMBER completed, but the node may still be a demoted voting member because promotion to voting member did not finish before attach\_voting\_member\_wait\_promotion\_timeout\_s expired. | Check the member status and allow promotion to complete. The attach operation still succeeds; do not blindly retry it. | | ER\_CONSENSUS\_LEADERSHIP\_TRANSFER\_TIMEOUT | Not available | 2754 | 2753 | Leadership transfer timed out while detaching or removing the current MA. Leadership transfer is an internal step, not a standalone user command. | Check the target voting member and quorum, then retry the DETACH or REMOVE operation. | | ER\_CONSENSUS\_LEADERSHIP\_TRANSFER\_FAILED | Not available | 2755 | 2754 | Leadership could not be transferred while detaching or removing the current MA. | Check the voting-member and quorum state, resolve the underlying issue, then retry the DETACH or REMOVE operation. | Monitoring election activity - Use `INFORMATION_SCHEMA.MV_EVENTS` to monitor MA election and consensus activity. Relevant events include `CONSENSUS_ELECTION`, `CONSENSUS_VOTING_EVENTS`, and `NEW_MASTER_AGGREGATOR`. *** Modified at: October 1, 2026 Source: [/db/v9.1/user-and-cluster-administration/high-availability-and-disaster-recovery/managing-high-availability/master-aggregator-redundancy-error-codes/](https://docs.singlestore.com/db/v9.1/user-and-cluster-administration/high-availability-and-disaster-recovery/managing-high-availability/master-aggregator-redundancy-error-codes/) (An index of the documentation is available at /llms.txt)