Bugzilla

Socorro was written a long time ago in a world with different standards. We're moving to a metrics standard where the key is prefixed with the service name. This is in addition to setting the `app` tag.

Antenna does this wrong. The application code emits metrics like "breakpad_resource.on_post" and the infrastructure sets up telegraf to prefix all keys emitted by Antenna with "socorro.collector" by setting the statds_namespace.

Socorro processor, webapp, and crontabber do this wrong. The application code emits metrics like "processor.es.ESCrashStorage.save_processed_crash".

We want to fix both of these in the GCP migration such that the application emits the complete key and telegraf doesn't make any changes to it.

For example, Antenna code would emit "socorro.collector.breakpad_resource.on_post".

For example, Socorro processor code would emit "socorro.processor.es.ESCrashStorage.save_processed_crash".

We want to make sure these things are in effect in GCP, but not AWS.

Socorro was written a long time ago in a world with different standards. We're moving to a metrics standard where the key is prefixed with the service name. This is in addition to setting the `app` tag.

Antenna does this wrong. The application code emits metrics like "breakpad_resource.on_post" and the infrastructure sets up telegraf to prefix all keys emitted by Antenna with "socorro.collector" by setting the statds_namespace.

Socorro processor, webapp, and crontabber do this wrong. The application code emits metrics like "processor.es.save_processed_crash".

We want to fix both of these in the GCP migration such that the application emits the complete key and telegraf doesn't make any changes to it.

For example, Antenna code would emit "socorro.collector.breakpad_resource.on_post".

For example, Socorro processor code would emit "socorro.processor.es.save_processed_crash".

We want to make sure these things are in effect in GCP, but not AWS.

Socorro was written a long time ago in a world with different standards. We're moving to a metrics standard where the key is prefixed with the service name. This is in addition to setting the `app` tag.

Antenna does this wrong. The application code emits metrics like `breakpad_resource.on_post` and there's a `STATSD_NAMESPACE` configuration variable which is passed into Markus `DatadogBackend` which prefixes all emitted keys with `socorro.collector`.

Socorro processor, webapp, and crontabber do this wrong. The application code emits metrics like `processor.es.save_processed_crash` and nothing adds the prefix.

We want to fix both of these in the GCP migration such that the application emits the complete key, we're not setting a `statsd_namespace`, and telegraf doesn't make any changes to it.

For example, Antenna code running in GCP would emit "socorro.collector.breakpad_resource.on_post".

For example, Socorro processor code running in GCP would emit "socorro.processor.es.save_processed_crash".

Also, we want to make sure these things are in effect in GCP, but not AWS. For AWS, we want it to run like it's currently running.

Socorro was written a long time ago in a world with different standards. We're moving to a metrics standard where the key is prefixed with the service name. This is in addition to setting the `app` tag.

Antenna does this wrong. The application code emits metrics like `breakpad_resource.on_post` and there's a `STATSD_NAMESPACE` configuration variable which is passed into Markus `DatadogBackend` which prefixes all emitted keys with `socorro.collector`.

Socorro processor, webapp, and crontabber do this wrong. The application code emits metrics like `processor.es.save_processed_crash` and nothing adds the prefix.

We want to fix both of these in the GCP migration such that the application emits the complete key, we're not setting a `statsd_namespace`, and telegraf doesn't make any changes to it.

For example, Antenna code running in GCP would emit `socorro.collector.breakpad_resource.on_post`.

For example, Socorro processor code running in GCP would emit `socorro.processor.es.save_processed_crash`.

Also, we want to make sure these things are in effect in GCP, but not AWS. For AWS, we want it to run like it's currently running.