I was just wondering if there was an established threshold at which web V4 on Synchronet became more "civilized" in its behavior. Don't want to throw a bunch of RAM in this thing just to find out that the cpu clock speed is below what is needed or that I could have avoided the mess just by adjusting a swappiness parameter.
know of any particular reason why ecWebv4 would be any slower than the legacy/runemaster web interface, but I'm sure it's fixable if we (or more specifically, echicken) had details (e.g. log messages, specific reproduction steps, etc.).
Likely the way message threads are displayed, vs a single message... probably a full scan of message base looking for downstream replies.
On 16 Jun 2023, Tracker1 said the following...
Likely the way message threads are displayed, vs a single message... probably a full scan of message base looking for downstream replies.
i duno, i tried vert.synchro.net via http and the initial load, clicking the forum link, clicking an area, and then a thread.. all those clicks take almost exactly 10 seconds
you'd think it'd be more performant than that. does ssjs have a profiler to see where it's spending all that time? there's not a sleep(10000) in there somewhere right? :-)
If you want a better example, use web.synchro.net instead. Same ecWeb
(and Synchronet) code, but different OS (vert is running on Win10 while web.synchro.net is running on Linux).
sheesh the difference is insane.. th
can't just be windows can it?
On 16 Jun 2023, Digital Man said the following...
If you want a better example, use web.synchro.net instead. Same ecWeb (and Synchronet) code, but different OS (vert is running on Win10 while web.synchro.net is running on Linux).
sheesh the difference is insane.. that can't just be windows can it?
Tracker1 wrote to Weatherman <=-
Re: Re: Web Server
By: Weatherman to Nightfox on Sun May 14 2023 01:55:00
I was just wondering if there was an established threshold at which web V4 on Synchronet became more "civilized" in its behavior. Don't want to throw a bunch of RAM in this thing just to find out that the cpu clock speed is below what is needed or that I could have avoided the mess just by adjusting a swappiness parameter.
Old message, my my biggeest suggestion would be to use an SSD or NVME drive if you can. You don't necessarily need to make a huge investment
in ram, the VM I'm running SBBS on via Docker has 4GB, and most of the time isn't even using 1/4 of that. The disk scanning for messages can
be slow on some systems/drives. Also, if you're on Linux, you can
check your settings for open file handles, etc. There's usually advice around this for "Linux File Server" as a search term, even though your application is different.
Windows might just need more resources. My BBS is running on Linux and the web interface is slow like vert (vs web.) but it only has 1.5gb ram.
Sysop: | Keyop |
---|---|
Location: | Huddersfield, West Yorkshire, UK |
Users: | 418 |
Nodes: | 16 (0 / 16) |
Uptime: | 14:38:03 |
Calls: | 8,795 |
Calls today: | 7 |
Files: | 13,298 |
Messages: | 5,966,560 |