Dmitri Bourlatchkov
Dmitri Bourlatchkov
I guess we have to provide explicit `Cache-Control` headers. It might work for `HEAD` as well, if we set the max age appropriately. With a reasonably low max age caching...
Actually, I meant to try and run two (or more) test jobs for _each_ backend (C* 3.11, 4.0, DSE) to as to reduce test time even more.
Will use the standard authZ checks that apply across all entity types for now.
> Setting cdc_enabled: true in cassandra.yaml should ensure that cdc param ends up in the schema, but that might actually enable CDC on Stargate nodes Interesting :thinking: I suppose enabling...
@marksurnin : Thanks for the update! I started investigating this in more depth from our side. Meanwhile, I noticed that schema handling has been refactored in C* 3.11.10 and 4.0-beta4....
It looks like the handling of CDC flags is not 100% robust even in plain C*. Here's my test case with 3.11.8: 1. Create and start a two node cluster...
@marksurnin : If I understand correctly, in your env. CDC is disabled and enabled is different DCs, right? I'd guess the above schema disagreement scenario might be possible in your...
@marksurnin : I suppose it may be possible for Stargate to poll C* storage nodes for schema data from its own code, then collect schema mutations and apply those that...
Thanks for the update @marksurnin ! Looking at C* code I get the impression that a lot of the logic around sending or not sending CDC flags was added to...
@pratyakshsharma : feel fee :+1: Please note that the docs might have already been fixed in many places since this issue was filed.