Hi I am trying to setup a sensor so whenever there is a fail

Question

Hi I am trying to setup a sensor so whenever there is a failure in the job rerun the same job with different offset I am using the run failure sensor for one of the job but when their is a failure it is not rerunning that same job could someone help me on this

Jordan · Answer

Hi probably you forgot to activate the sensor you can do it by going to your instance on dagit or add `default status = DefaultSensorStatus RUNNING` in your sensor definition to have the default behaviour

Manga Dhatrika · Answer

ok how could I pass the dynamic offset at each failure what I am trying to achieve is I have suppose 1000 records that I am extracting from the API in first run it extracted 502 records and failed I would like to pass in a parameter from the sensor where it will get me the records that was loaded means 502 and when it reruns it sends the offset 503 and extract from 503 records and load

Manga Dhatrika · Answer

instead of rerunning the whole job again

Manga Dhatrika · Answer

this is code I am using it still did not rerun the job ` run failure sensor ` `job selection= extract and load data default status = DefaultSensorStatus RUNNING so the sensor only runs when this job fails` ` ` `def rerun db fetch sensor context RunFailureSensorContext ` ` logic to get the rows to start on` `yield RunRequest ` `run config= ` `job name= extract and load data ` ` `

Manga Dhatrika · Answer

in dasgter daemon I see `2022 05 23 16 39 33 0400 dagster daemon SensorDaemon INFO Completed a reaction request for run 0b6ec3dd 7095 4664 8711 2ed2168108d2 Sensor rerun db fetch sensor acted on run status FAILURE of run 0b6ec3dd 7095 4664 8711 2ed2168108d2 `

Manga Dhatrika · Answer

I am new to dagster so I dont know what I am doing wrong

sean · Answer

Hi Manga Sorry for the late response here give me a minute to look into this for you

sean · Answer

So it turns out that `run failure sensor` currently does not support yielding `RunRequest` like other sensors We are aware of this issue and that our docs are confusing and have an issue for it <https github com dagster io dagster issues 7383> You can expect this to be corrected in the next few weeks