move the schema migration out of collector

[srikanthccv]

Running multiple replicas of collectors creates issues, as no atomic lock migration is in place. Some potential options

1. Use a standalone migration repo/image
2. Use the init scripts of ClickHouse

[nityanandagohain]

With multiple clickhouse ditry can be resoved by deleting both schema_migrations and logs_attribute_keys table.

[ankitnayan]

Use the init scripts of ClickHouse

we do not need to restart clickhouse for migrations to apply. If we use init scripts of clickhouse, we would have to restart clickhouse whenever we apply migrations

https://andrewlock.net/deploying-asp-net-core-applications-to-kubernetes-part-7-running-database-migrations/

[ELLIOTTCABLE]

Oof, yeah, being unable to safely run multiple collectors in parallel is a bit of an issue for us.

I think, for the moment, then, our only option is to run signoz-otel-collector as a 'thin', ingest-only instance on a single box; and have our multiple non-forked otel-collector instances feed that "thin" ingestor over OTLP HTTP or something; have the non-forked instances perform any batching/aggregation/transformation … hrm.

[ankitnayan]

When we move out, we can also look to support templating of cluster name as requested in #180

[ankitnayan]

@ELLIOTTCABLE this is in progress and should be released soon