Conversation

Notices

1. cassidyclown (cassidyclown@clubcyberia.co)'s status on Friday, 23-Feb-2024 00:44:57 JST cassidyclown
   finally got us a server non-fraudulently we're back cyberians
   •  (mint@ryona.agency)'s status on Friday, 23-Feb-2024 00:44:56 JST 

     in reply to
     @cassidyclown >non-fraudulently
     >not putting tyrone mcniggerman as legal name
     >not providing them with a fake ID scan in some latam/easteuro shithole in case they have concerns
     >not asking some kid on fiverr to shake his head on camera if even that fails
   • Edward Sneeden (sneeden@iddqd.social)'s status on Friday, 23-Feb-2024 00:52:19 JST Edward Sneeden

     in reply to

     • 
     @mint @cassidyclown writing this down
      likes this.
   •  (mint@ryona.agency)'s status on Friday, 23-Feb-2024 00:55:38 JST 

     in reply to

     • pwm
     • Phantasm
     @phnt @cassidyclown @pwm Smarr indie hosting company, prease understandu.
   • Phantasm (phnt@fluffytail.org)'s status on Friday, 23-Feb-2024 00:55:39 JST Phantasm

     in reply to

     • pwm
     • Phantasm
     @cassidyclown @pwm Also if anything breaks on the VPS side, it's probably BuyVM's fault. One time they had two instances of the same VM running. Other time they literally force powered it off and never started it back up.
     
     https://fluffytail.org/notice/Aa3u6k0nbZA8Pn434i
   • Phantasm (phnt@fluffytail.org)'s status on Friday, 23-Feb-2024 00:55:40 JST Phantasm

     in reply to

     • pwm
     @cassidyclown @pwm It works, just don't put the DB on it. I've seen few people talk about long iowait when under heavy load.
   • pwm (pwm@crlf.ninja)'s status on Friday, 23-Feb-2024 00:55:41 JST pwm

     in reply to
     @cassidyclown
   • cassidyclown (cassidyclown@clubcyberia.co)'s status on Friday, 23-Feb-2024 00:55:41 JST cassidyclown

     in reply to

     • pwm
     @pwm that block storage better perform or I'm going to kill him for making me turn my VPN off
   • Phantasm (phnt@fluffytail.org)'s status on Friday, 23-Feb-2024 00:57:49 JST Phantasm

     in reply to

     • 
     • pwm
     @mint @cassidyclown @pwm Wouldn't choose anybody else. The pricing is almost unbeatable for what you get.
      likes this.
   • 0 (https://gh0st.live/users/0)'s status on Friday, 23-Feb-2024 01:01:41 JST 0

     in reply to

     • pwm
     @cassidyclown @pwm
     
     Block storage is like emergency usb tier response. It's not the same as local disk at all.
   •  (mint@ryona.agency)'s status on Friday, 23-Feb-2024 01:01:41 JST 

     in reply to

     • pwm
     • 0
     @0 @cassidyclown @pwm I hosted all the instance media on it, and it was fine. Guess it is suitable as a data warehouse where things are mostly read, not written.
   • pwm (pwm@crlf.ninja)'s status on Friday, 23-Feb-2024 01:02:18 JST pwm

     in reply to

     • 
     • 0
     @mint @cassidyclown @0 what do you bet it's smr under there?
      likes this.
   •  (mint@ryona.agency)'s status on Friday, 23-Feb-2024 01:02:59 JST 

     in reply to

     • pwm
     • 0
     @pwm @cassidyclown @0 Definely spinning rust, don't remember if I've done any benchmarks.
   •  (mint@ryona.agency)'s status on Friday, 23-Feb-2024 01:08:11 JST 

     in reply to

     • 
     • pwm
     • 0
     @pwm @0 @cassidyclown Wrote a gigabyte file with dd both on the root filesystem and on a storage slab. 800 MB/s on root, ~500-600 MB/s on slab. Too lazy to do proper benchmarks, so I don't know what to gather from those results.
   • 7666 (7666@comp.lain.la)'s status on Friday, 23-Feb-2024 01:33:30 JST 7666

     in reply to

     • 
     • pwm
     • 0
     @mint @cassidyclown @pwm @0 I use it as a 3TB cache for all cached files and doing direct reads and writes to it was downright awful.
     
     I decided to make it into a LVM PV, wrote a thin allocated file on the local SSD storage, turned that into a loopback device, and enabled LVM cache using that file. It's okayish now with the cache in front.
      likes this.
   •  (mint@ryona.agency)'s status on Friday, 23-Feb-2024 01:34:24 JST 

     in reply to

     • 7666
     • pwm
     • 0
     @7666 @cassidyclown @pwm @0 Funny that I wanted to do the same, but decided against it after benchmarks.
   • Phantasm (phnt@fluffytail.org)'s status on Friday, 23-Feb-2024 01:34:48 JST Phantasm

     in reply to

     • 
     • pwm
     • 0
     @mint @cassidyclown @pwm @0 It's fine with sequential r/w, random access kills it completely.
      likes this.
   • :blank: (i@declin.eu)'s status on Friday, 23-Feb-2024 01:56:33 JST :blank:

     in reply to

     • 
     • 7666
     • pwm
     • 0
     @7666 @cassidyclown @pwm @0 @mint discord does the same thing on GCP to get latency down, but it would probably still work well enough if postgres was told it had limited parallel io and random pages cost 4x the usual amount
     
     still a shame pleroma fails so badly on a slow disk, crashing on a loop
      likes this.