as the developer and operator of the server she is interacting with, the problem is between YOUR keyboard and YOUR wall at this moment in your technical support diagnostics, i have already confirmed that my relay is not storing parameterized replaceable events at all, at all, by backing up its whole database and filtering it for "kind":30040 which she also reported seemed to upload previously
thanks for your service
that was referring to the computer as their DNS is failing
ah, i missed that haha, apologies
if the address was valid then the fault was between their pc and the DNS server they are using, not the person behind the keyboard
anyway, nostr:npub1l5sga6xg72phsz5422ykujprejwud075ggrr3z2hwyrfgr7eylqstegx9z maybe you should visit your computer's DNS configuration and choose another DNS to use... ISPs are notoriously crappy, that's why i almost invariably use cloudflare or google DNS
I can however confirm I did move the relay from my test environment and didn’t move the data, and there was some downtime
also I am considering using a good amount of the realy code, which looks good, just with a different event store and some other changes
will contribute back those changes if possible
one thing you may want to do is separate the actual DB logic and what I call the index provider
that basically allows you to swap the query planner and index generation logic, and allows making search very easy (tokenize => save as individual index entries)
index changes become easy too, as all you need to do is reindex all events in place, with no interruptions, and after that’s done enable the new indexes
I can literally do that in a one liner with realy, pull the events as JSONL from one relay and direct the stdout from that over to the import endpoint of the other relay, with zero downtime
i've not actually tried that, but i use curl routinely to pull and push events over the http-basic auth protected admin endpoint i built in a couple months ago
bragging aside tho, my relay doesn't handle 30k numbered events at all haha
