When the application is leaking connections, the only way to recover is to restart the whole application. We need instead to automatically recover the pool once the connections are detected as leaked.
- add a marker on the connection for having an active time to live
- state and kill connection which have timed out periodically
- add a counter and metric about killed active connections
- add an operation for invoking the cleanup by hand
- depends on
-
NXP-19541 validationQuery is not taken in account by the connection pool
- Resolved
-
NXP-19724 backport validation query fixup in 7.10
- Resolved
- is related to
-
NXP-21229 Conditional ignore rule should take in account annotated methods
- Resolved
- is required by
-
NXP-26623 Deactivate connection killer by default
- Resolved