- Support traffic replay as an experimental feature #642
- Make the API
/api/debug/healthreturn error when no backends are available #692 - Do not log config when the config is not changed #694
- Fix that the configuration
proxyis reset when the command line flag--advertise-addris set #691 - Fix that an unhealthy backend may never be removed #697
- Change the tiproxyctl flag
--curlsto--hostand--port#664
- Support virtual IP management #583
- Read metrics from backends for load balance when Prometheus is unavailable #465
- Fix the bug that backend metrics are not cleared after backends are down #585
- Fix vulnerability issues scanned by third-party tools #623
- Support multi-factor-based balance #465
- Add configuration
proxy.advertise-addr#495 - Add command line argument
--advertise-addr#497 - Speed up health check by checking in parallel #498
- Make sure the Welcome TiProxy info is always printed #507
- Do not reject new connections during
graceful-wait-before-shutdown#525 - Add configuration
labels#536 - Set the Y axis of some metrics to
logBase=2#561
- Deprecate command line arguments
log_levelandlog_encoder#504
- Replace
time.Timewith mono-time to optimize duration calculation #461 - Change the log level of some logs from info to debug #463
- Optimize updating metrics to improve the performance by around 3% #467
- Reduce GC CPU usage by 1% in the case of plenty of connections #474
- Add metrics for traffic and handshake #477
- Add alert rules #481
- Support get JSON format config for
/api/admin/configby settingAccept='application/json'in the request header #484
- Fix that welcome logs may not printed #454
- Fix that rebalance may not work after session migrations are interrupted #459
- Fix the bug that the config file may not be reloaded if the directory is removed and recreated immediately #475
- Fix panic when a TiDB fails during session migration #486
- Fix the bug that some goroutines are not recovered after they panic #488
- Support online reload for
require-backend-tlsand remove unnecessary configs #396 - Add config
graceful-close-conn-timeout#400 - Add config
server-http-tlsto specify HTTP port TLS config #403 - Do not report errors when the sequence mismatches #410
- Double-check the target backend health before session migration #412
- Add more metrics for connections, queries and health checks #416
- Move config
require-backend-tlsfromproxytosecurityand change the default value tofalse#419 - Add HTTP API
/debugand/metricsto be consistent with TiDB #426 - Remove pushing metrics to Prometheus to avoid authentication #431
- Change the default minimum TLS version from v1.1 to v1.2 #437