“inode” in Linux
Inode — a term most people have never heard before. Even the experienced DevOps engineers are not aware of this inode term. Whenever I talk about Linux commands with some DevOps folks I know, they don’t know what an inode is, and I hear that “they’ve never worked on it.”
But the fun part is, “Everyone who has worked with the Linux terminal or the file structure has worked with inodes.”
An inode is a small record that stores the metadata about the file or folder in the Linux file system. It stores the metadata of a file and folder, like: File size, Owner (user/group), Permissions (rwx), Timestamps (created, modified, accessed), File type (file, folder, link, device), Pointers to where the file’s actual data is stored on disk
To be honest, I’ve been using Linux since my college days; it’s been around four years now, and even though I never really worked on it, until last week
An incident when I realised how important understanding INODE is:
Suddenly, my team's messages started popping up, saying the pipeline is failing for the projects one after another. The Gitlab Runner is installed on the same server where all the production applications are deployed, and we were getting a “No space left on server ” error in the pipeline
Get Aniket Shitole’s stories in your inbox
Join Medium for free to get updates from this writer.
When I went inside the server and checked what was taking up too much space, I saw that the space is not occupied by anything; we have plenty of space
When I dug deeper, I found that there are countless log files. They were not consuming the space, but each one of them had an inode, and that’s why my server was running out of inode numbers
As you can imagine, I had no time to take a screenshot at that time, and no one would want to make inode numbers full in the server just to take a screenshot 😂
How I Fixed the “Inode Full” Issue and Got Everything Running Again
Once I realised that the disk space was fine and the real problem was the inode count hitting 100%, the next step was to find where all these tiny files were hiding.
In my case, it pointed to something like: /temp/logs
I found a logs folder and deleted it completely, and finally, the inode could breathe.
After deleting those files, when I ran $df -i I could see plenty of inodes free
And with that, the server came back to life. CI/CD pipelines stopped failing, and the “No space left” error disappeared
This incident taught me that even the smallest components in Linux, like inodes, can bring an entire production system down if ignored
I love sharing the knowledge I’ve gained so others don’t face the same challenges I encountered. I write about topics like DevOps, Linux, AWS, and more. Follow me for more helpful insights and tips! 🙏
