1
0
mirror of https://github.com/jbranchaud/til synced 2026-01-19 06:58:02 +00:00
Files
til/rails/clean-up-memory-hungry-rails-console-processes.md

44 lines
2.2 KiB
Markdown

# Clean Up Memory Hungry Rails Console Processes
I noticed (using `htop`) that a remote server hosting a Rails app had most of
its RAM being actively consumed. This was hindering my ability to run a fresh
deploy because the deploy processes had to do a ton of memory swapping which
drastically slowed the whole thing down.
With some investigation, I discovered that most of the memory was being consumed
by a handful of `rails console` processes. I didn't have any known active `rails console` processes that I was using. That combined with the dates of these
processes starting way in the past suggested to me that these were abandoned
processes that hadn't been properly cleaned up.
```bash
server:~# ps aux | grep rails
32767 878915 0.0 0.0 1227160 936 pts/0 Ssl+ 2025 0:03 /exec rails console
32767 878942 0.9 6.5 830996 261748 pts/0 Rl+ 2025 249:51 ruby /app/bin/rails console
32767 3004097 0.0 0.0 1227160 692 pts/0 Ssl+ 2025 0:04 /exec rails console
32767 3004129 0.9 6.4 834672 257228 pts/0 Dl+ 2025 406:31 ruby /app/bin/rails console
32767 3048582 0.0 0.0 1227160 940 pts/0 Ssl+ Jan09 0:00 /exec rails console
32767 3048611 1.1 6.3 829936 253484 pts/0 Dl+ Jan09 60:50 ruby /app/bin/rails console
32767 3060033 0.0 0.0 1227160 944 pts/0 Ssl+ 2025 0:04 /exec rails console
32767 3060063 0.9 6.5 838084 260812 pts/0 Rl+ 2025 405:37 ruby /app/bin/rails console
root 3699372 0.0 0.0 7008 1300 pts/0 S+ 15:51 0:00 grep --color=auto rails
server:~# ps aux | grep 'rails console' | awk '{sum+=$6} END {print sum/1024 " MB"}'
1014.64 MB
```
As we can see by tacking on this `awk` command, these processes are consuming
1GB of memory.
Each of these is a pair of processes. A parent process (`/exec rails console`)
that kicks off and supervises the memory-hungry child process (`ruby /app/bin/rails console`).
To free up this memory, I targeted each of the parent processes with a `kill`
command one by one. For example:
```bash
server:~# kill 878915
```
I suspect that I may have left the occasional terminal tab open with one of
these `rails console` processes running and the SSH connection was getting
killed without the `rails console` getting killed with it.