mirror of
https://github.com/mailscope/kumomta.git
synced 2026-08-18 18:38:18 +00:00
docs: update for release
This commit is contained in:
@@ -0,0 +1,37 @@
|
||||
# Release 2025.01.29-833f82a8
|
||||
|
||||
## Other Changes and Enhancements
|
||||
|
||||
* Added [kcli inspect-sched-q](../reference/kcli/inspect-sched-q.md) command. #231
|
||||
* Added
|
||||
[reconnect_strategy](../reference/kumo/make_egress_path/reconnect_strategy.md)
|
||||
egress path option to control what happens with a session that experiences
|
||||
a disconnection during message sending.
|
||||
* Improved robustness of [kcli
|
||||
trace-smtp-server](../reference/kcli/trace-smtp-server.md) and [kcli
|
||||
trace-smtp-client](../reference/kcli/trace-smtp-client.md) on busy servers.
|
||||
* kafka client can now process non-UTF8 binary payloads. Thanks to @cai-n! #328
|
||||
* Exposed the `max_burst` throttle parameter to the throttle parser. This
|
||||
functionality was always present, it just wasn't configurable. See the
|
||||
[kumo.make_throttle](../reference/kumo/make_throttle.md) docs for the revised
|
||||
throttle specification.
|
||||
|
||||
## Fixes
|
||||
|
||||
* Regression with the recent RSET optimizations: we didn't issue an RSET if a send
|
||||
failed partway through, leading to issues with the connection state.
|
||||
* SMTP Client could sometimes get stuck attempting to process a series of messages
|
||||
on a connection that had previously been closed.
|
||||
* Potential cache thrashing issue with `remember_broken_tls` could lead to a larger
|
||||
number of connection attempts to sites with broken TLS.
|
||||
* The RFC3464 parser could fail to parse certain `Arrival-Date` headers. Thanks
|
||||
to @cai-n! #329
|
||||
* The lua context pool would not invalidate entries when the [config
|
||||
epoch](../reference/configuration.md#config-epoch) changed. This meant that
|
||||
if you had config values encoded in the lua script itself (rather than via a
|
||||
memoized load from a datasource), you would need to wait until the [maximum
|
||||
lua context age](../reference/kumo/set_max_lua_context_age.md) had expired
|
||||
for those changes to be reflected.
|
||||
* Changing the composition (sources) of an egress pool might not fully take
|
||||
effect until the associated queues had aged out.
|
||||
|
||||
@@ -4,35 +4,5 @@
|
||||
|
||||
## Other Changes and Enhancements
|
||||
|
||||
* Added [kcli inspect-sched-q](../reference/kcli/inspect-sched-q.md) command. #231
|
||||
* Added
|
||||
[reconnect_strategy](../reference/kumo/make_egress_path/reconnect_strategy.md)
|
||||
egress path option to control what happens with a session that experiences
|
||||
a disconnection during message sending.
|
||||
* Improved robustness of [kcli
|
||||
trace-smtp-server](../reference/kcli/trace-smtp-server.md) and [kcli
|
||||
trace-smtp-client](../reference/kcli/trace-smtp-client.md) on busy servers.
|
||||
* kafka client can now process non-UTF8 binary payloads. Thanks to @cai-n! #328
|
||||
* Exposed the `max_burst` throttle parameter to the throttle parser. This
|
||||
functionality was always present, it just wasn't configurable. See the
|
||||
[kumo.make_throttle](../reference/kumo/make_throttle.md) docs for the revised
|
||||
throttle specification.
|
||||
|
||||
## Fixes
|
||||
|
||||
* Regression with the recent RSET optimizations: we didn't issue an RSET if a send
|
||||
failed partway through, leading to issues with the connection state.
|
||||
* SMTP Client could sometimes get stuck attempting to process a series of messages
|
||||
on a connection that had previously been closed.
|
||||
* Potential cache thrashing issue with `remember_broken_tls` could lead to a larger
|
||||
number of connection attempts to sites with broken TLS.
|
||||
* The RFC3464 parser could fail to parse certain `Arrival-Date` headers. Thanks
|
||||
to @cai-n! #329
|
||||
* The lua context pool would not invalidate entries when the [config
|
||||
epoch](../reference/configuration.md#config-epoch) changed. This meant that
|
||||
if you had config values encoded in the lua script itself (rather than via a
|
||||
memoized load from a datasource), you would need to wait until the [maximum
|
||||
lua context age](../reference/kumo/set_max_lua_context_age.md) had expired
|
||||
for those changes to be reflected.
|
||||
* Changing the composition (sources) of an egress pool might not fully take
|
||||
effect until the associated queues had aged out.
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
# reconnect_strategy
|
||||
|
||||
{{since('dev')}}
|
||||
{{since('2025.01.29-833f82a8')}}
|
||||
|
||||
Controls the behavior of the SMTP dispatcher when it encounters an error during
|
||||
message delivery. It does **not** affect connection-time errors (such as connection
|
||||
|
||||
@@ -52,7 +52,7 @@ where quantity is a number and period can be a measure of time.
|
||||
of that one minute interval. If you set `6/m,max_burst=1` then the rate at
|
||||
which they can proceed will be 1 every 10 seconds (*interval* = *period* /
|
||||
*quantity*, so: `60/6 == 10`) across that minute. Larger burst values will
|
||||
allow larger bursts across that time period. {{since('dev', inline=True)}}
|
||||
allow larger bursts across that time period. {{since('2025.01.29-833f82a8', inline=True)}}
|
||||
|
||||
Examples of throttles:
|
||||
|
||||
|
||||
Reference in New Issue
Block a user