I am using ERPNext in Digital Ocean with 512 ram. After installing few days, database is crashing frequently by itself and requires manual starting. This installation is in test mode, so there are no traffic or heavy usage! I tried to reduce innodb_buffer_pool to 128MB from 256MB, but did not work! Can anyone please advice that how I can make it right?
What OS? This seems pretty low… but you can try running free -m while the db is running and see what you have left. I think you probably need to bump this up… traffic or no, httpd, mariadb, supervisor, celery… that’s more than you have in 512.
@mar I am running ERPNext on DO with CentOS 7 and 2GB Ram with no issues. I suggest you allocate more RAM.
Failing this In CentOS 7 you can create a “swap” which allows you to allocate part of your HD as virtual RAM. This of course increases latency massively but prevents OOM errors. I guess Ubuntu has the same feature. You can read more here:
In short allocate more RAM.
Also make sure you have a minimal install or whatever server OS you use. If errors persist delete current instance and rebuild from scratch.
2 Likes
With 512 MB … yes you will run into trouble.
Read the DO page where they explain how to make / use your SWAP memory. Create a swap memory of 1G and it should run fast.
Best,
François
Its has been few days I created swap, and everything is running fine. Thank you everyone. 
2 Likes