what is the difference between --es-archive.* and --es.*

[yuecong]

I saw almost the same settings exist both for --es-archive.* and --es.*. Could someone explain what the difference between these two? Thanks

[jpkrohling]

@pavolloffay can give you more info, but this is for the case where you have your recent data in one place (mostly writes), and older data (mostly read-only) in another ES index/server, for performance reasons.

I realized that we might not have info about this in the docs, so, I'll leave this item open to track the task of properly documenting this.

[yuecong]

Thanks for the info. awesome. btw, I think people may use hor-warm architecture so that it can be one ES cluster to handle both read/write.

[pavolloffay]

The --es-archive.* or --cassandra-archive are used to store archive traces. There is a button in the UI in trace detail view which stores a trace to archive storage.

I have realized that we do not document the archive storage anywhere. At least I could not find any docs. @yurishkuro are there any docs explaining archive storage?

I am moving this to website/docs repository

Btw. hot-warm indices should be even used for normal daily indices. Where only the currrent daily index would be kept in the hot nodes.

[pavolloffay]

This docs could go to deployment or https://www.jaegertracing.io/docs/1.14/cli/

[yurishkuro]

CLI strings could use a bit of improvement like mentioning the storage role a given option refers to (although it's kind of obvious from the option name).

But I think the main issue is we need to describe what archive storage is and why it's used in the other docs, maybe Architecture section.