Datadog logging improvements
J
James Brewer
A couple of things that would improve the logs being sent to Datadog:
* "syslog.appname" and "syslog.hostname" should be swapped. "syslog.appname" maps to the "service" attribute in Datadog, which shouldn't changed based on the number of instances being run. "syslog.hostname" maps to the "host" attribute in Datadog, which should change based on the number of instances being run.
* Some logs that should be "info" are showing up as "alert". For example, Sidekiq logs about a job completing include "lvl: 'INFO'" in the log, but are showing up in Datadog as "alert".
Log In
A
Ankit Shah
I would like to also see these improvements in DataDog logging. It would be also good to actually set
source
as well. With
source
, it would be safer to then add additional transformations in DataDog as pipeline. https://docs.datadoghq.com/logs/log_configuration/pipelines/?tab=source#overviewM
Matthew Maclean
Any ability to configure Datadog related reserved attributes (source/service/env/etc) would significantly improve the logging experience. See https://docs.datadoghq.com/integrations/postgres/?tab=host.
In addition, multiline support would be nice.
J
James Brewer
Postgres logs use the database name as the "service" attribute in Datadog instead of the Render service name. E.g. We have a service called "company-postgres-staging" with the database name set to "company_staging_xpga". We'd prefer that "company-postgres-staging" be the service name in Datadog.