Got a huge spike in Memory yesterday
I am looking at my billing and saw one of my instances https://railway.app/project/cccc72dc-c205-4c79-80cb-eb2f054ee9d1/service/d6626c5d-8907-490c-a60f-cae2a38d0878/metrics had a huge spike in memory usage yesterday.
What goes into this? Do logs effect my memory usage?
Railway
Railway
Railway is an infrastructure platform where you can provision infrastructure, develop with that infrastructure locally, and then deploy to the cloud.
20 Replies
Project ID:
cccc72dc-c205-4c79-80cb-eb2f054ee9d1,d6626c5d-8907-490c-a60f-cae2a38d0878define huge, and do you see the same memory spike in the service metrics
the CPU looked fine but the memory spiked from 300mb to 1.3gb
does the service metrics reflect this spike
i guess im just trying to understand why my bill went from 3bucks to 9bucks this month idk
let me look at the billing again
i am reading the chart on the metrics page ya one sec
can you show me a zoomed in view, maybe set to 1day
I guess just generally do you have any recommendations to reduce my memory?
Sure one sec let me grab that
this was the last day
okay spike only visable in the 7d settings
when the spike happened, did it fall back down before any new deployments?
ahh like a mem leak
thats a good q
yeah thats what im getting at
there shouldn't be a mem leak on anything other than logs
we have a set amount of cache that we cache on the server
only 1 endpoint caches calls
ya i guess i ll monitor the mem change before my next long gap in deployments
thanks
console logs to stdout/stderr should have absloely no impact on mem
do you use some type of logging agent?
no i don't
interesting
the app is pretty simple
- a couple routes talk to my database
- a route that talks to a test blockchain node
It's possible that at spikes there is a lot of information returned from my blockchain and the memory hasn't been freed yet ?
It's possible that at spikes there is a lot of information returned from my blockchain and the memory hasn't been freed yet ?
nodejs?
yup
express app
memory leaks are very possible with node, but did the mem usage return to normal levels after the spike? becaucse it would only be classifiable as a mem leak if the mem didnt return to normal levels
ya i guess i am not sure
next time i deploy i will check the current mem usage
to see if that is what is causing this to return to normal
sounds good