This commit adds upgrade and downgrade hooks that are called when
upgrading from a bridge V1 to connector and action or the other way
around. The automatic translation is used if the callback is not
defined.
NOTE: Backported from master
This commit fixes the upgrade and downgrade strategy when upgrading
from a bridge V1 to connector and action or the other way around so that
the custom callbacks get the complete unchanged input instead of the
result of the automatic translation. The automatic translation is used
if the callback is not defined.
This commit changes how the `emqx_action_info` callback
`bridge_v1_to_action_fixup/1` works. It is now called after the
automatic upgrade instead of before. Since the full Bridge V1 config
might be needed to do the fixup, it is provided in a special field
`<<"__bridge_v1_conf__">>`. The `<<"__bridge_v1_conf__">>` field is
removed after the callback is called and can thus be ignored if it is
not needed.
This commit adds callbacks to the emqx_action_info module for
doing fixes (such as changing a field name) after the automatic
split of a Bridge V1 config or the merge of connector and action
configs for the compatibility layer.
Fixes https://emqx.atlassian.net/browse/EMQX-11330
After feedback from Product team, we should rename `bridges_v2` to `actions` everywhere.
We'll start with the public facing APIs.
- HTTP API
- Hocon schema root key
Even though most of the time these modules will be used by
connectors, there are exceptions (namely, `emqx_rule_engine`).
Besides, they are general enough to land there, more so given
that `emqx_placeholder` is already there.
Treat "${$}" as literal "$". This allows to template express
strings, for example, of the form "${some_var_value}" where
`some_var_value` is interpolated from bindings.
Missing bindings in string templates will be rendered as "undefined",
as before. Rendering still assumes that missing binding with implicit
default (`undefined`) is an error.
This will also restore complete backward compat in `emqx_prometheus`.
This slightly changes semantics: now the attempt to create authenticator
with illegal bindings in templates will fail, instead of treating them
as literals. The runtime behaviour on the other hand should be the same.
Located under `emqx_connector` for now. The APIs of the new modules
are slightly different from the old ones. The new ones are more
explicit in terms of error handling. Also copy the according testsuite
from `emqx_plugin_libs` and enrich it for better coverage.
Co-authored-by: Thales Macedo Garitezi <thalesmg@gmail.com>
Co-authored-by: Stefan Strigler <stefan.strigler@emqx.io>
Co-authored-by: Zaiming (Stone) Shi <zmstone@gmail.com>
Several bridges should be able to share a connector pool defined by a
single connector. The connectors should be possible to enable and
disable similar to how one can disable and enable bridges. There should
also be an API for checking the status of a connector and for
add/edit/delete connectors similar to the current bridge API.
Issues:
https://emqx.atlassian.net/browse/EMQX-10805
Fixes https://emqx.atlassian.net/browse/EMQX-10629
During health checking, we check whether tables in the SQL statement
exist. Such check was done by asking the backend to parse the
statement using a named prepared statements. Concurrent health checks
could then result in the error:
```erlang
{error,{error,error,<<"42P05">>,duplicate_prepared_statement,<<"prepared statement \"get_status\" already exists">>,[{file,<<"prepare.c">>},{line,<<"451">>},{routine,<<"StorePreparedStatement">>},{severity,<<"ERROR">>}]}}
```
This could lead to an inconsistent state in the driver process, which
would crash later when a message from the backend (`READY_FOR_QUERY`, "idle"):
```
2023-07-24T13:05:58.892043+00:00 [error] Generic server <0.2134.0> terminating. Reason: {'module could not be loaded',[{undefined,handle_message,[90,<<"I">>,...
```
Added calls to `epgsql:sync/1` for functions that could return
`{error, sync_required}`.
Also, redundant calls to `parse2` were removed to reduce the number of requests.
Fixes https://emqx.atlassian.net/browse/EMQX-10405
The problem here was that, for async requests, ehttpc responses of the form `{ok, 4__, _,
_}` and similar were being treated as successes.
Fixes https://emqx.atlassian.net/browse/EMQX-9603
Rather than relying on a JWT worker to produce and refresh tokens, we
could just produce then on demand when pushing the messages to GCP
PubSub. That can generate a bit of extra work (as multiple processes
might realize it’s time to refresh the JWT and do so), but that
shouldn’t be much. In return, we avoid any possibility of not having
a fresh JWT when pushing messages.
That currently tunes the number of MQTT clients employed both for
subscriptions (if shared subscription is used) and for publishing to
a remote broker.
It adds no value: the only mode was `cluster_shareload` and we just as
well can decide to "share" the load across cluster just by looking if
the remote topic is shared subcription filter or not.
Inline `emqx_connector_mqtt_msg` module code into
`emqx_connector_mqtt_worker` module, since it's not really used
anywhere else and does not provide any reusable abstractions.
Thales Macedo Garitezi