ulimit: Setting Resource Limits for Processes

Applications fail with 'too many open files' or 'cannot fork' errors at scale because default OS resource limits are too low.

Increase ulimits in the shell for testing, and persistently in /etc/security/limits.conf for production.

# View current limits
ulimit -a

# View specific limits
ulimit -n   # open file descriptors
ulimit -u   # max user processes
ulimit -s   # stack size
ulimit -m   # max memory

# Set open file limit for current session
ulimit -n 65536

# For persistence: /etc/security/limits.conf
# username  soft  nofile  65536
# username  hard  nofile  131072
# *         soft  nproc   4096

# For systemd services, set in unit file:
# [Service]
# LimitNOFILE=65536
# LimitNPROC=4096

# Check actual limits for a running process
cat /proc/$(pgrep nginx | head -1)/limits

# System-wide limit
sysctl fs.file-max
sysctl -w fs.file-max=2097152

ulimits have two levels: soft (current effective limit, user can raise up to hard) and hard (ceiling set by root). Services like databases, web servers, and Elasticsearch commonly need nofile raised to 65536 or higher.