All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Suspended again for high CPU usage
averagedatahoarder
Member
Hey guys, I know how this sounds, this is not a complaint against any provider. I know my CPU usage was excessive, it looks like I had either a memory leak or runaway process from my logging/analytics system. Either way, it resulted in a VPS suspension due to excessive resource usage.
This was probably a fair suspension given the TOS regarding CPU usage tbh. But I just don't know what to do at this point. This is a free API I host with over a billion requests a month and (only ~5-10TB bandwidth usage, don't worry, it's a different service) - but I just don't know where to host it. If I get a VPS how do I know this won't happen again? I don't need the full resources of a VDS, I just need something reliable with decent uptime.
I know everyone probably thinks this is a joke or something but I am genuinely trying to find a solution. I also do apologize to all the hosting providers I may have offended in any way.
Well, now I'll ask -- does anyone have recommendations for me?
My current plan is:
XEON 16G
4 Cores
16GB RAM
100GB Storage
6.99€ per month
Anything similar would be greatly appreciated, doesn't even need to be unmetered I guess, I just care about storage (ideally 150-200GB would be excellent) and at least 15-25TB of bandwidth for some buffer as I ramp this up. Also a better CPU would be nice.
I'm willing to pay a bit extra, I just need something with good reliability.
Anyways, if anyone has suggestions please let me know.
Comments
nice.
P.S. Maybe @layer7 but it seems like you should get dedi resources contrary to
XEON is too weak. You need to use Ryzen.
buy dedi, if you want dedicated hw
based on your usage you need to have dedicated CPU, so you never get suspension for high CPU usage
Yeah I know, just much more expensive. And I don't need full CPU usage this was due to a bug/mistake in my code
for 10 euros get kimsufi and call it a day.
the 10 euro kimsufis are not available yet
servarica and advinservers seem to be popular
Or maybe an ovh vps? Not sure about the CPU on those though.
Please show API documentation, source code, and deployment instructions (exact command to deploy the software).
We can then audit them and give suggestions on how to prevent excessive CPU usage again.
Servarica sold out
@Advin do you offer disk upgrades?
In this AI times you need 2 min to find a way to get one..
You really willing to do this?
Servarica is available, but prices have bumped up recently for new orders.
Never heard of them before, but vds.net is advertising on the sidebar.
knowing @averagedatahoarder , he'd probably get suspended for using too much CPU on a dedi too..
@SKRIME help a man in need
I'm on skrime rn btw
I know
Thanks for the tag @nghialele.
@averagedatahoarder thanks for being upfront about it. Once you've found the runaway process, open a ticket and we'll unsuspend the server.
If you want more headroom, you can also upgrade through a ticket and move to EPYC with more cores. The EPYC 24G gives you 6 cores on AMD EPYC Zen 3 Milan, 24 GB RAM and 150 GB NVMe for 18.99€/mo. If you'd like to stay at 16 GB, the configurator does 8 cores, 16 GB and 100 GB for 19.89€/mo. More cores means a single runaway process can't take the whole server with it.
Either way, put a hard cap on the logging and analytics part. If it runs as a systemd service, add
CPUQuota=200%andMemoryMax=2Gunder[Service], plusRestart=on-failureso it comes back after a leak gets it killed. In Docker the same thing is--cpus=2 --memory=2g.Thanks, I opened a ticket. I guess I just don't want this to happen again and run into these issues. But, 6.99 euros to 18.99 is a big jump
More like too much electricity then
Assuming
CPUQuota=200%referred to the 4 vcores, does that mean your FUP for CPU is 50%😬
suspended for gooning
double it and pass it to the next person
Wait a bit, they are preparing their next sale currently. Looking at all your posts, you'll love Kimsufis and they are most likely your best option long term.
Ty
What kind of API is it?
have you considered just not doing that
@sumo no, the 200% was just an example value for capping that one logging service, it has nothing to do with our rules. The fair use is about a server running at 100% permanently, and in practice that's the only case where we step in. The cap is only there so a single buggy process can't take the whole server with it, so pick whatever value fits your setup.