Usage/cost spike
I have a issue with the usage of my service suddenly skyrocketing, despite my "bots" not being used.
My cost did a 10x suddenly.
Happy to pay the invoice if there is any reason for the sudden spike in usage.
I checked the logs which displays the usage each time the bot is called, and don't see any change from normal.
I am mainly running some discord bots.
56 Replies
Project ID:
4a08fc5a-47e8-49d4-bc9f-411ad640ab0bwhat resource usage has skyrocketed
here are the project ids:
- 4a08fc5a-47e8-49d4-bc9f-411ad640ab0b
- 2677e6f2-094d-43c5-b3f3-4e94db2b566d
- 2aecae83-d13d-4f3c-949b-816e2244b888
I don't work for Railway so those unfortunately don't help me
here is the current usage.
however my average cost per month was always rouhgly 3-4$
how many projects do you have?
3 projects.
how many of those have excessive costs?
4 instances running total
all three of them
have you looked at the cost per service breakdown for the project?
yes all look like the memory was the cause of the spike, despite having barely any actual usage. bots are mostly inactive usage wise
maybe a memory leak?
do you have a bunch of environments on that project?
enviroment variables you mean?
nope
project environments
oh no not really total of 4
across all 3 projects
I think you're talking about services, so I'll take that as a no for the project environment question
what kind of worker is this?
so I only host discord bots and one twitter bot (currently not even working due to updated API terms from X)
what language?
3 wokers in JS and 1 in Py
what do the service metrics say
those are project metrics, please look into your service metrics
oh ok sorry one sec
and what language is this one
JS
also JS
nothing wrong with that one
also JS
also nothing wrong
Py - this one isn't even working atm since the twitter APIs aren't active anymore
that's a decent amount of egress
but all workers barely get any usage. so how die the first worker I showed above, manage to do 1.4gb in memory
that could be the reason for high costs on that service
it looks like it dropped back off, switch to 6h and send another screenshot
instantly drops back down
looks good now, what was the cause of the drop? did you redeploy?
nop nothing, I rarely touch these bots.
I didn't even push anything to GIT
2 months ago was the last git update
well unfortunately there's really nothing railway could tell you here, railway runs your code as is after all
but I don't see any logs showing usage
I log each incoming trigger to monitor the usage, and I don't see any reason for ALL my services to have a spike in usage
yes this is extremely odd, but I'm sorry that doesn't change the statement I made above
as for the service with the high network usage, does it connect to a database?
So I run a service, there is no actual usage, I go from 0-3$ in monthly usage to 33$
all three instances barely get used, and my logs store each trigger, and I don't see ANY trigger from either the bots, to justify the usage cost.
nop
railway bills for the resources used, if your app uses 1.4gb of ram then that's what they bill for
so they technically csn bill me whatever and I have to just accept?
it's what your app has used
despite having no proof that my instances used the ressources
buy my app haven’t been used that the thing
what you app has used is in that graph
railway runs your code as-is
can I at least talk to a railway staff? I would like to look further into this as I cannot justify the usage! I understand you comment, but please understand my point here also. We run a instance for almost a year with avg. cost of 0-3$ a month. Our bot usage haven’t spiked YET our cost has. So I would like to look further into this issue.
a hobby user would only get community support
the team would only be brought in when platform issues are involved
but there is a platform issue. My metric don’t match my usage. And I can’t contact them because I can’t upgrade my account
I am not paying for the usage until I find out WHY my projects have a sudden spike with no correlation to real world usage. My cost did a 10x without a single user using my applications which before cost me ca. 3$ a month on average accross all projects for a period of almost 1year. So WHY suddenly do I get a 10x price increase, with no real world usage increase (which is verifiable via my logs, which each time triggers a entry when a user calls the bot)
Not trying to get out of paying, that isn’t the issue here. I purely want to understand what happened, because it wasnt due to usage if you ask me!
I just don't see how railway could give you the answer you seek, it's your code and your code was ran as-is
I think we are turning in circle. A code that was running for 11months non stop generates 3$ a month worth of invoice based on the usage. Now fast forward the present, the same code, no changes and no increase in usage from end-users, and my cost do a 10x.
How can that be related to my code? I mean it doesn’t suddenly go rogue and do calls on its own. At the end of the day, my bots run on a trigger and each time the bot gets triggered it uses processing power other wise it sits in idle mode. Now I log EVERY single request from a user to analyse how many requests we get. And based on those logs, my usage remains on paar with the months that cost me 3$, so why now the 33$ invoice. Again I am not wanting to talk myself out of paying the invoice, just not a fan of not understanding what id going on.
Again, from my understanding there is something weird going on, and I don’t think it is unreasonable to ask those questions to a company that invoices me.
is there at least an email I can contact?
team support is only for pro users, trial / hobby users get community support
just wanna say ive seen my fair share of support requets where the user said the same thing you are saying "my app / code didnt use these resources" where the teams answer was (in short) "yes it did"
I understand and not saying I am right, but I don’t think it is a healthy business praxis to not have a line of communication open for people that have concerns over their invoices. I am not being unreasonable with my concerns, and did my research to evaluate if the spike in service cost is justified or not. Based on that, I came to conclusion that … no it isn’t justified and therefore looking for an explanation.
Since there is not direct support here on discord and as you mentioned no other ways to reach out to the team, I won’t waste you time further and will look at another way to reach out to the team.
Thank you for your support anyways and appreciate your time ✨