TiCDC Alert Rules
This document describes the TiCDC alert rules and the corresponding solutions. In descending order, the severity levels are: Critical, Warning.
Critical alerts
This section introduces critical alerts and solutions.
cdc_checkpoint_high_delay
For critical alerts, you need to pay close attention to abnormal monitoring metrics.
Alert rule:
ticdc_owner_checkpoint_ts_lag > 600Description:
A replication task is delayed more than 10 minutes.
Solution:
cdc_resolvedts_high_delay
Alert rule:
ticdc_owner_resolved_ts_lag > 300Description:
The Resolved TS of a replication task is delayed more than 5 minutes.
Solution:
ticdc_changefeed_failed
Alert rule:
(max_over_time(ticdc_owner_status[1m]) == 2) > 0Description:
A replication task encounters an unrecoverable error and enters the failed state.
Solution:
This alert is similar to replication interruption. See TiCDC Handles Replication Interruption.
ticdc_processor_exit_with_error_count
Alert rule:
changes(ticdc_processor_exit_with_error_count[1m]) > 0Description:
A replication task reports an error and exits.
Solution:
Warning alerts
Warning alerts are a reminder for an issue or error.
cdc_multiple_owners
Alert rule:
sum(rate(ticdc_owner_ownership_counter[30s])) >= 2Description:
There are multiple owners in the TiCDC cluster.
Solution:
Collect TiCDC logs to locate the root cause.
cdc_sink_flush_duration_time_more_than_10s
Alert rule:
histogram_quantile(0.9, rate(ticdc_sink_txn_worker_flush_duration[1m])) > 10Description:
It takes a replication task more than 10 seconds to write data to the downstream database.
Solution:
Check whether there are problems in the downstream database.
cdc_processor_checkpoint_tso_no_change_for_1m
Alert rule:
changes(ticdc_processor_checkpoint_ts[1m]) < 1Description:
A replication task has not advanced for more than 1 minute.
Solution:
ticdc_puller_entry_sorter_sort_bucket
Alert rule:
histogram_quantile(0.9, rate(ticdc_puller_entry_sorter_sort_bucket{}[1m])) > 1Description:
The delay of TiCDC puller entry sorter is too high.
Solution:
Collect TiCDC logs to locate the root cause.
ticdc_puller_entry_sorter_merge_bucket
Alert rule:
histogram_quantile(0.9, rate(ticdc_puller_entry_sorter_merge_bucket{}[1m])) > 1Description:
The delay of TiCDC puller entry sorter merge is too high.
Solution:
Collect TiCDC logs to locate the root cause.
tikv_cdc_min_resolved_ts_no_change_for_1m
Alert rule:
changes(tikv_cdc_min_resolved_ts[1m]) < 1 and ON (instance) tikv_cdc_region_resolve_status{status="resolved"} > 0 and ON (instance) tikv_cdc_captured_region_total > 0Description:
The minimum Resolved TS 1 of TiKV CDC has not advanced for 1 minute.
Solution:
Collect TiKV logs to locate the root cause.
tikv_cdc_scan_duration_seconds_more_than_10min
Alert rule:
histogram_quantile(0.9, rate(tikv_cdc_scan_duration_seconds_bucket{}[1m])) > 600Description:
The TiKV CDC module has scanned for incremental replication for more than 10 minutes.
Solution:
Collect TiCDC monitoring metrics and TiKV logs to locate the root cause.
ticdc_sink_mysql_execution_error
Alert rule:
changes(ticdc_sink_mysql_execution_error[1m]) > 0Description:
An error occurs when a replication task writes data to the downstream MySQL.
Solution:
There are many possible root causes. See Troubleshoot TiCDC.
ticdc_memory_abnormal
Alert rule:
go_memstats_heap_alloc_bytes{job="ticdc"} > 1e+10Description:
The TiCDC heap memory usage exceeds 10 GiB.
Solution:
Collect TiCDC logs to locate the root cause.