Details
-
Improvement
-
Resolution: Fixed
-
Minor
-
master
-
Couchbase server
-
1
Description
Post an incident with a customer, when collecting logs via cbcollect_info, the only filesystem status collected is 'df -ha' as per the following line in the script:
UnixTask("Filesystem", "df -ha"), |
Linux filesystems have two content limits:
- Physical capacity, seen by the output of the `df -ha` command
- Logical capacity, seen by the amount of index-nodes (inodes), seen by the output of the `df -i` command
A file system is considered full, once it reaches to the minimum between the two limits mentioned above.
The inode status is currently not reflected in the cbcollect_info script, collecting the customer's environment information.
I propose to add the following command post the current 'df -ha' command in the cbcollect_info script:
UnixTask("Filesystem inodes", "df -i"), |
The amount of output is roughly the same as the current 'df -ha' command, but instead of the data blocks output provided for each mounted filesystem we will get its inode status.