Usage Memory Spikes
See attached image. I have 2 containers which together rarely exceed 800mb, but I'm seeing spikes to 2.8gb and 5+gb on the usage graph (presumably effecting the billing?). Can you help me understand how this is calculated?
22 Replies
Can't really tell sadly. We just run the code you give us.
We have a ticket to at some point add "hit rates"
But this is >> the sum of the mem utilization you report
I.e. what's the discrepancy?
ProjectID?
Could be multiple environments etc
Railway
Railway
Railway is an infrastructure platform where you can provision infrastructure, develop with that infrastructure locally, and then deploy to the cloud.
One environment, two apps
When did this get deployed?
22nd looks like
Interesting. I do see your point...
My only thought is it's either builds or simultaneous deployments. But having 5x seems crazy.
Could be simultaneous deployments
What drives billing?
You like, index something off the rip?
(When you start your app)
Nah these are simple webapps
Or do some sort of startup operation?
Sum of all deployed instances
Is that what's being reported on the usage page?
Ye
Need to make it more clear
Give you a lil drillable breakdown
If so would be great to have visibility to the delta
Yeah perfect
Completely immaterial ATM but when we're running a few grand on you guys soon it'll be more important
Yea for sure
Lemme bang drum about this at our triage meeting
Which, was Friday but got moved to Monday. So, i'll raise it internally
Will also dig in a bit and see if I can get an answer for you sooner
No rush, thanks
Roger and welcome!
Hey @nootrality - spent some time digging deeper as to why you are seeing that. Two parallel but related threads going on.
1) So, for some reason your CPU and memory are reporting for a tiny spike in usage wayyyyy above normal. This looks very shocking in the usage graph but balances out in the cost department when we normalize the usage data. (For the record, we log at 5 minute intervals and average out the usage.)
When looking at the cost calculation, nothing out of the normal appears, the $2 of usage so far seems correct.
2) The UX of the usage graph is confusing. I think we can do some color coding there to show which deployment is causing what cost. In the Project page we do have a per service breakdown but I think we can do better on the graphs to make that grokkable
https://railway.app/project/2db6fbc5-2ff6-48d5-b635-66e792767de5/settings/usage
I have spun up an escalation for 1) and 2) internally. We can't get to this immediately but I appreciate you giving us the information that you have
Let us know if there is anything else we can answer for you on the platform
this is great, thanks angelo!