ngrilly

44 ms·

1. 661.▲
   by ngrilly 7y ago

   What do you mean by "embedded database libraries"? Libraries like SQLite or RocksDB? What kind of issue are you thinking of?
2. 662.▲
   by ngrilly 7y ago

   Thanks! This probably means we can expect some IOPS and latency improvements when they'll upgrade :)
3. 663.▲
   by ngrilly 7y ago

   I read DigitalOcean is using Ceph for its object and block storage. Do they use BlueStore in production too?
4. 664.▲
   by ngrilly 7y ago

   Short TTLs are bad, but I confess using them in some small scale projects in anticipation of this kind of worst case scenario :)
5. 665.▲
   by ngrilly 7y ago

   Do your DNS records have a short TTL to enable such a quick switch from one provider to another?
6. 666.▲
   by ngrilly 7y ago

   Will you migrate back to Netlify when Zeit will suffer a similar downtime?
7. 667.▲
   by ngrilly 7y ago

   > It's encrypted with your keys (which I didn't see on Bitwarden's site) Same with Bitwarden: https://help.bitwarden.com/article/can-bitwarden-see-my-pass...
8. 668.▲
   by ngrilly 7y ago

   I can see what you mean, but most startups try to have users as early as possible, to demonstrate traction. Most also practice continuous deployment. In the early stage, a lot of refactoring usually happens, which often requires database mi
9. 669.▲
   by ngrilly 7y ago

   Nope. > SQLite4 was an experimental rewrite of SQLite that was active from 2012 through 2014. All development work on SQLite4 has ended. Lessons learned from SQLite4 have been folded into the main SQLite3 product. SQLite4 was never relea
10. 670.▲
    by ngrilly 7y ago

    Another big factor you don't mention is concurrency. SQLite is single writer/multiple readers, unlike PostgreSQL which supports multiple concurrent writers. This makes hard to run database migrations without blocking writes during
11. 671.▲
    by ngrilly 7y ago

    > I'd go as far as to say that many modern startups could get pretty far with SQLite + aggressive caching before even bringing in a big database like Postgres Modern startups need their services to be available, and this is difficul
12. 672.▲
    by ngrilly 7y ago

    I'm not sure you are measuring what you think you are measuring. Have you set CONN_MAX_AGE in Django?
13. 673.▲
    by ngrilly 7y ago

    Bad benchmark with not enough information to draw any relevant conclusion. My bet: Connection pooling is not enabled for PostgreSQL (Django default value for CONN_MAX_AGE is 0). A new connection is established for each request, which probab
14. 674.▲
    by ngrilly 7y ago

    Yes, DO uses Ceph: https://blog.digitalocean.com/why-we-chose-ceph-to-build-blo...
15. 675.▲
    by ngrilly 7y ago

    Since you're trying to "delete file with lifecycles", I'm quite sure your problem is with their object storage (called Spaces), and not their block storage.
16. 676.▲
    by ngrilly 7y ago

    Can't agree more. That said, Kubernetes still has way too much RAM/CPU overhead in a single node setup. It scales up, but it doesn't scale down. Hopefully, it can be solved. There is k3s, for example, that limits the overhead
17. 677.▲
    by ngrilly 7y ago

    Moving away to what?
18. 678.▲
    by ngrilly 7y ago

    Custom hardware? What kind and how is it related to Borg?
19. 679.▲
    by ngrilly 7y ago

    On the surface, it looks similar to Telegram groups. What are the differences?
20. 680.▲
    by ngrilly 7y ago

    Ugh! Was the fix committed upstream?
21. 681.▲
    by ngrilly 7y ago

    I haven't noticed that.
22. 682.▲
    by ngrilly 7y ago

    Yes, it's a real issue. Kubernetes burns too much RAM and CPU even with no workload.
23. 683.▲
    by ngrilly 7y ago

    I know. I was writing about the 1:1 model without async/await :)
24. 684.▲
    by ngrilly 7y ago

    > blocking I/O Rust I/O are strictly similar to Go when using 1:1 OS threading, but become conceptually different when using async/await. > Rust has more features for safe concurrency--Rust prevents data races staticall
25. 685.▲
    by ngrilly 7y ago

    Rust is really great, but I feel like you are overselling it a bit here: - async/await is available in the nightly version of Rust (1.39), but not in the stable version (1.37). - 1:1 threading is not strictly equivalent to M:N threadin
26. 686.▲
    by ngrilly 7y ago

    > they even scale well on Linux to all but the absolute most demanding workloads What about the stack size? I guess the default is about 4 MB, compared to a few kB in Erlang or Rust. This implies a virtual memory usage 100x or 1000x high
27. 687.▲
    by ngrilly 7y ago

    > In Java, they'll behave like async/await, only without the syntactic constructs. Stacks and stack frames are moved, can shrink and grow, can have various representations etc., and the scheduling is entirely pluggable. You can
28. 688.▲
    by ngrilly 7y ago

    This is very interesting! I'd be happy to try. My main pain points with Telegram are: - No way to structure a group into sub-groups - No way to comment on a specific post without polluting the whole timeline (the new DiscussionBot is t
29. 689.▲
    by ngrilly 7y ago

    Thanks!
30. 690.▲
    by ngrilly 7y ago

    Which channel?