Commit Graph
16 Commits
Author SHA1 Message Date
Jan Losinski d0b9f08bf2 InfluxDB send retry after IOError (#10263)
* Implement data write retry for InfluxDB

This adds an optional max_retries parameter to the InfluxDB component
to specify if and how often the component should try to send the data
if the connection failed due to an IOError.

The sending will be scheduled for a retry in 20 seconds as often as the
user specified. This can be handy for flaky getwork connections between
the DB and Homeassistant or outages like daily DSL reconnects.

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Add unittest for influx write retries

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Add RetryOnError as helper decorator in util

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Add unittests for RetryOnError

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Use RetryOnError decorator in InfluxDB

This replaces the scheduling logic in the InfluxDB component with the
RetryOnError decorator from homeassistant.util

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Make the linters happy

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Implement a queue limit for the retry decorator.

This adds a queue limit to the RetryOnError handler. It limits the
number of calls waiting for be retried. If this number is exceeded,
every new call will discard the oldest one in the queue.

* influxdb: Add the retry queue limit option.

* Make the linter happy.

* Make pylint happy

* Log exception of dropped retry

* Move RetryOnError decorator to influxdb component.

* Fix bug in logging usage

* Fix imports

* Add newlines at the end of files.

* Remove blank line

* Remove blank line
2017-11-23 16:58:18 -08:00
Jan Losinski 693d32fa68 Snapcast: bump version and enable reconnect. (#10626)
This bumps the used snapcast version to 2.0.8 and enables the new
reconnect feature that causes the component to reconnect to a server if
the connection was lost.

This fixes the ned to restart Home Assstant after a snapcast reboot, as
described in issue #10264.

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>
2017-11-16 20:32:26 -05:00
Jan Losinski fd6fd765b2 Pilight switch: restore last state after restart (#8580)
* Pilight switch: restore last state after restart

This uses the restore_state helper to set the last known state to
pilight switches when the devices are initialized after a HA
restart.

Without this HA forget the state on every restart and needs to be told
the sttae by retoggling the switches. This can cause unwanted effects
as a switch toggling may emit an RF signal.

* Make hound happy

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Remove entity_id generation as requested in review.

* Make hound happy again.

* fix comments

* fix lint
2017-08-22 16:40:14 +02:00
Jan Losinski fde4a7d029 Citybikes: Allow None as result for empty slots (#8528)
The citibykes API returns "null" as value for empty_slots on some
stations (see #8527). This causes the component to not process the data.

This is fixed by accepting None as valid data. The row in the frontend
is left empty if "null" was returned by the service.

fixes #8527
2017-07-17 22:50:55 +02:00
Jan Losinski 3374169c74 Allow InfluxDB to blacklist domains (#7264)
* Allow InfluxDB to blacklist domains

This adds an option to InfluxDB to blacklist whole domains. This is
useful for domains like automation or script, where no statistic data
is needed.

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Add unittest for InfluxDB domain blacklist

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Use common include/exclude config for InfluxDB.

Its now the same syntax as it is for recorder.

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Add unittests for InfluxDB include whitelist.

There where no tests for that feature before.

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>
2017-04-26 21:14:52 +02:00
Jan Losinski 215987d5a7 Add script to import state events to InfluxDB (#7254)
* Add script to import state events to InfluxDB

This adds a script to import recorded events from a recorder database
to a InfluxDB instance. This can be useful for initial importing after
setup of a InfluxDB.

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Fix step argument handling in Influx import

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Correct typo in InfluxDB Importer

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Update influxdb_import.py
2017-04-24 23:01:09 +02:00
Jan Losinski c32300a386 Bump limitlessled dependency to 1.0.5. (#6334)
This fixes issue #6295.
2017-03-02 14:18:44 +01:00
Jan Losinski 41218e5a37 [switch.pilight] Implement echo config option (#5056)
* Implement echo config option for pilight

Pilight switches can get a receive code configured. If so they act on
received codes. In the current implementation "act on" means not only
to set the internal state, but also to send the code again. If the
receiver is directly parred with the switch, to act even if HA is not
running, this causes it to receive the signal twice because the HA
sends it again.

In my case this causes a dimmer to start dimming until I hit the switch
again.

This implements a "echo" argument for the receive codes that let the
user choose if a received signal should result in any sending or not.
If not, only the status of pilight will be updated.

The echo option defaults to True, as this was the default since now.
Simply set it to halse to disable this behaviour.

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Add documentation to set_state in switch/pilight.

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>
2017-01-26 22:40:14 -08:00
Jan Losinski 469aad5fc8 Add teardown method to pilight tests (#5195)
* Add teardown method to pilight tests

This is necessary to stop the HomeAssistant instance that was started
in the setUp method. Without this there happen random test failures.

This is necessary to stabilize the tests for PR #5045.

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Update test_pilight.py
2017-01-08 14:23:01 +01:00
Jan Losinski f78246e686 Pilight receive match fix for bug 4637 (#4639)
* Pilight: dont protocol as list in COMMAND_SCHEMA

As described in bug #4637 the protocol should not be wrapped in a list
in the spec of COMMAND_SCHEMA because this causes the component to
never successfully match any received rf code.

As pointed ot in PR #4639 the easiest way to do this, is to not derive
COMMAND_SCHEMA from RF_CODE_SCHEMA and specify protocol as simple
string there.

This fixes bug #4637.

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Pilight: Add "unitcode" to command schema.

This adds "unitcode" to the COMMAND_SCHEMA. It is used for example in
the brennenstuhl protocol of pilight.

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>
2016-12-08 23:19:14 -08:00
Jan Losinski d0dcd1bb73 Scene: add support for input_select (#4674)
This adds support for the scene component to handle input_select
devices and set their options.

This fixes bug #4673

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>
2016-12-07 05:33:41 -08:00
Jan Losinski 5c807c6bd9 MPD: Reconnect mpd client afetr OSError (#4651)
If the mpd client ran into an socket timeout, the socket will raise an
OSError on every further request. This adds OSError to the list of
excptions, that causes a client reconnect.

This fixes #4650

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>
2016-12-01 12:28:31 -08:00
Jan Losinski dd84b4e237 Mpd: Use "file" instead "id" for media_content_id (#4653)
In media_content_id() the "id" of the current song was returned. as
stated in bug #4652 the id is only the Tracklist-Id in the current
tracklist and is omitted if the track is not part of a tracklist (what
caused the bug in the first place).

To match the semantics described in the dockstring, to return a "Content
ID", this chooses the filename of the current song as id and returns
it.

It also uses get() instead of [] to prevent KeyError.

This fixes bug #4652

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>
2016-12-01 12:20:42 -08:00
Jan Losinski 200bdb30ff Change pilight systemcode validation to integer (#4286)
* Change pilight systemcode validation to integer

According to the pilight code the systemcode should be an integer and
not a string (it is an int in the pilight code). Passing this as a
string caused errors from pilight:
"ERROR: elro_800_switch: insufficient number of arguments"

This fixes #4282

* Change pilight unit-id to positive integer

According to the pilight code the unit of an entity is also evrywhere
handled as an integer. So converting and passing this as string causes
pilight not to work.

This fixes #4282

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>
2016-11-10 21:41:28 -08:00
Jan Losinski 7d2ab4fce6 Change pilight systemcode validation to integer (#4286)
* Change pilight systemcode validation to integer

According to the pilight code the systemcode should be an integer and
not a string (it is an int in the pilight code). Passing this as a
string caused errors from pilight:
"ERROR: elro_800_switch: insufficient number of arguments"

This fixes #4282

* Change pilight unit-id to positive integer

According to the pilight code the unit of an entity is also evrywhere
handled as an integer. So converting and passing this as string causes
pilight not to work.

This fixes #4282

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>
2016-11-10 13:14:40 -08:00
Jan Losinski 52eb816c62 Introduce a send_delay for pilight component (#4051)
* Add a method to throttle calls to services

This adds CallRateDelayThrottle. This is a class that provides an
decorator to throttle calls to services. Instead of the Throttle in
homeassistant.util it does this by delaying all subsequent calls
instead of just dropping them. Dropping of calls would be bad if we
call services to actual change the state of a connected hardware (like
rf controlled power plugs).

Ihe delay is done by rescheduling the call using
track_point_in_utc_time from homeassistant.helpers.event so it should
not block the mainloop at all.

* Add unittests for CallRateDelayThrottle

Signed-off-by: Jan Losinski <losinski@wh2.tu-dresden.de>

* Introduce a send_delay for pilight component

If pilight is used with a "pilight USB Nano" between the daemon and the
hardware, we must use a delay between sending multiple signals.
Otherwise the hardware will just skip random codes. We hit this
condition for example, if we switch a group of pilight switches on or
off. Without the delay, random switch signals will not be transmitted by
the RF transmitter.

As this seems not necessary, if the transmitter is directly connected
via GPIO, we introduce a optional configuration to set the delay.

* Add unittests for pilight send_delay handling

This adds an unittest to test the delayed calls to the send_code
service.
2016-11-01 21:50:27 -07:00