When fly.io introduced swap_size_mb, I moved from creating the swap manually with a script to letting fly handle swap creation
And since then I’m getting the following error during fly deploy
#16 110.8 apps/***.com build: <--- Last few GCs --->
#16 110.8 apps/***.com build: [226:0x69db440] 79655 ms: Mark-Compact (reduce) 2039.4 (2081.6) -> 2039.3 (2082.1) MB, 1327.93 / 0.01 ms (+ 34.2 ms in 6 steps since start of marking, biggest step 17.1 ms, walltime since start of marking 1385 ms) (average mu = 0.304, current mu = 0.2[226:0x69db440] 81531 ms: Mark-Compact (reduce) 2039.7 (2082.3) -> 2039.7 (2082.6) MB, 1359.31 / 0.00 ms (average mu = 0.290, current mu = 0.275) allocation failure; GC in old space requested
#16 110.8 apps/***.com build: <--- JS stacktrace --->
#16 110.8 apps/***.com build: FATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memory
#16 110.8 apps/***.com build: ----- Native stack trace -----
#16 110.8 apps/***.com build: 1: 0xb84bd6 node::OOMErrorHandler(char const*, v8::OOMDetails const&) [node]
#16 110.8 apps/***.com build: 2: 0xefec30 v8::Utils::ReportOOMFailure(v8::internal::Isolate*, char const*, v8::OOMDetails const&) [node]
#16 110.8 apps/***.com build: 3: 0xefef17 v8::internal::V8::FatalProcessOutOfMemory(v8::internal::Isolate*, char const*, v8::OOMDetails const&) [node]
#16 110.8 apps/***.com build: 4: 0x1110925 [node]
#16 110.8 apps/***.com build: 5: 0x11287a8 v8::internal::Heap::CollectGarbage(v8::internal::AllocationSpace, v8::internal::GarbageCollectionReason, v8::GCCallbackFlags) [node]
#16 110.8 apps/***.com build: 6: 0x10fe8c1 v8::internal::HeapAllocator::AllocateRawWithLightRetrySlowPath(int, v8::internal::AllocationType, v8::internal::AllocationOrigin, v8::internal::AllocationAlignment) [node]
#16 110.9 apps/***.com build: 7: 0x10ffa55 v8::internal::HeapAllocator::AllocateRawWithRetryOrFailSlowPath(int, v8::internal::AllocationType, v8::internal::AllocationOrigin, v8::internal::AllocationAlignment) [node]
#16 110.9 apps/***.com build: 8: 0x10dd0a6 v8::internal::Factory::NewFillerObject(int, v8::internal::AllocationAlignment, v8::internal::AllocationType, v8::internal::AllocationOrigin) [node]
#16 110.9 apps/***.com build: 9: 0x1537fe1 v8::internal::Runtime_AllocateInOldGeneration(int, unsigned long*, v8::internal::Isolate*) [node]
#16 110.9 apps/***.com build: 10: 0x7fc57fe99ef6
#16 111.2 apps/***.com build: Aborted (core dumped)
#16 111.2 apps/***.com build: ELIFECYCLE Command failed with exit code 134.
#16 111.2 apps/***.com build: ERROR: "build:remix" exited with 1.
so basically, before I was using the following script to create the swap manually at container boot time
#!/usr/bin/env node
import { writeFile } from 'node:fs/promises'
import { $ } from 'execa'
console.log('setting up swapfile...')
await $`fallocate -l 512M /swapfile`
await $`chmod 0600 /swapfile`
await $`mkswap /swapfile`
await writeFile('/proc/sys/vm/swappiness', '10')
await $`swapon /swapfile`
await writeFile('/proc/sys/vm/overcommit_memory', '1')
console.log('swapfile setup complete')
and now I ditched the script to instead use swap_size_mb inside my fly.toml file
the changes I applied match this commit dffa1e6
My project is not that large, 2 libraries and 1 app at the moment, in a monorepo managed with pnpm.
This error is happening while pnpm build is running inside the docker
Increasing the swap to 1024 doesn’t change a thing
Is someone else experiencing the same issue? How can I resolve this issue?