The z-index of the header was set to 2000 because of Leaflet JS zoom
controls going over the header instead of under when scrolling. However,
this caused issues with the notifications and dropdowns (menu and
autocomplete).
Setting the z-index to 1001 is enough to fix the Leaflet JS issues,
without causing other issues.
Wrapping the handling of integration events in a tokio::spawn should
already have been there, as we do not want to delay the downlink in case
of slow integrations.
This fixes the FrmPayload decryption in case of frame-counter rollover
(16lsb) as it was using the f_cnt as sent over the air (16lsb) and not
the full frame-counter (32b).
Before, these functions would return the device-session for the given
uplink PhyPayload (if a matching device-session was found), together
with the full frame-counter. However it would not modify the f_cnt of
the PhyPayload to the full frame-counter making it prone to errors like
the above.
The lorawan-devices repository structure is going to change and the
latest revisions no longer contain a LICENSE file.
This does mean that the latest data can't be imported and we will be
missing newly added devices and potential bugfixes. However, it does
provide time to work on a better solution.
As requested, TTN and The Things Network is no longer used in the
naming and description.
Closes#61.
redis::pipe() can't be used with the ClusterClient struct, instead we
must use cluster_pipe() to start the pipeline. This implements a wrapper
which constructs the pipeline based on the used Redis setup.
It can be useful to handle uplinks that do not match a DevEUI in a
separate process. This includes DevAddr to DevEUI pointer does not
exist, or DevAddr points to one or multiple DevEUIs, but it does not
pass the MIC check.
This makes it possible for external services to subscribe (through
Redis) for realtime events. E.g. a create, update or delete device event
could trigger an external synchronization.
In case there is a loadbalancer between ChirpStack and the MQTT broker,
it is useful to have control over the keep-alive messages, to make sure
that the loadbalancer does not drop the connection if there is no
activity. In such a case you want to set the keep_alive_interval to a
lower interval than the loadbalancer timeout.
In case the queue-item does fit the first RX window, but not the second
one, it will be set to None. If this happens, then the
save_downlink_frame step will not add the queue-item ID to the
DownlinkFrame (internal).
Then on txack, ChirpStack is unable to dequeue the payload causing the
payload to stay in the queue.
Per Regional Parameters Specification, the LoRa modulation always uses
coding-rate 4/5, however this is not the case for ISM2400 (which is
currently not covered by the reg. params) in which case coding-rate
4/8LI must be used.
Closes#51.
This fixes the issue where after a from DST to non-DST, the timstamp was
not incremented by a full day. As the ts was then truncated to day
precision, this would create a never ending loop.
Example, incrementing 2022-10-30 00:00:00 by days(1) would result in
2022-10-30 23:00:00, trucated to day precision would result in
2022-10-30 00:00:00.
If the date (30) is the same after incrementing by days(1), we increment
by days(2), which in the above example results in 2022-10-31 23:00:00,
truncated to day precision results in 2022-10-31 00:00:00 which is the
desired result.
In case the decoded payload contains random keys the auto-detect
measurements feature will add new measurements for each uplink. With
this option it is possible to turn auto-detection off to avoid
polluting the database with many measurements.
Closes#42.
This makes it possible to configure one or multiple DevAddr sub-ranges
instead of using the whole DevAddr range provided by the configured
NetID.
Closes#49.
This type defines the (NetID derived) prefix that must be used when
generating device addresses. This can be retrieved from the NetID, but
it is also possible to define a prefix that defines a smaller address
range within the NetID.
See also #49.