6 Common Linux Ops Issues and How to Fix Them Quickly

General Troubleshooting Process for Linux Operations Engineers

As a qualified Linux operations engineer, you need a clear fault‑resolution workflow: pay attention to error messages, examine log files, analyze and locate the cause, then fix the problem.

Problem 1 – Filesystem corruption prevents boot

When the boot log shows “/dev/sda6 contains a file system with errors, check forced”, the likely cause is an unexpected power loss leading to inconsistent filesystem structures. Fix it with umount /dev/sda6 and fsck.ext3 -y /dev/sda6.

Problem 2 – “Argument list too long” error

The error occurs when too many arguments are passed to a command, often because /var is full. Check with df -h, identify large directories (e.g., /var/spool/clientmqueue), and delete or clean them using patterns, find, or a shell script. If necessary, increase the kernel limit by editing include/linux/binfmts.h and recompiling the kernel.

Problem 3 – Inode exhaustion causing application failure

Even when disk space appears sufficient, running out of inodes prevents file creation. Verify with df -i and locate offending files (e.g., mail queues). Remove them with find /var/spool/clientmqueue/ -name "*" -exec rm -rf {} \; or inspect individual inode numbers using ls -i and stat.

Problem 4 – Deleted file does not free space

After deleting a large log file such as /tmp/access_log, space may remain occupied because a process (e.g., httpd) still holds the file descriptor. Use lsof | grep delete to find the locked file, then either stop the process or truncate the file with echo "" > /tmp/access_log to release space.

Problem 5 – “Too many open files” error

The error indicates that the per‑process file‑descriptor limit has been reached. Check the limit with ulimit -n and ensure the limit is applied in user profiles, startup scripts, or /etc/security/limits.conf. If the limit was added after the service started, restart the service (e.g., Tomcat) to apply the new limits.

Problem 6 – Read‑only filesystem

A read‑only mount usually results from filesystem corruption or hardware failure. Unmount the filesystem ( umount /www/data), stop any processes using it ( fuser -m /dev/sdb1 then ps -ef | grep 8800), run fsck -V -a /dev/sdb1, and remount.