The current StorageGRID implementation runs on the Novell SLES operating system. The design philosophy dedicates the root partition (
/) to relatively static content, that is, the operating system, configuration files, and StorageGRID software. Since the storage requirements are fairly constant, the StorageGRID servers were historically installed with a minimal space allocated for the root partition. The majority of volatile content is placed on large dedicated file systems. For example, the
/var/local/ partition contains almost all the logs and is generally used to store the temporary files during maintenance and support engagements.
It is common for older servers (especially on older grids) to have only 5 GB dedicated for their root partition. A hardware refresh usually results in a larger root file system.
The following three common scenarios can deplete the free space of the root partition:
- A third-party software (usually supplied by the hardware vendor) is installed and by default logs in to a directory other than
/var/local/. This is often the
/var/logdirectory or one of its sub-folders.
- An unusual issue is experienced by a user, which causes the normal logs to begin growing.
- The temporary files are placed on the wrong file system. This can be done easily, since the default directory at login is the
/root/ homedirectory. The
/tmpdirectory is sometimes a location for unnecessary or required data. Ensure to remove the data or copy the data to the
/var/local/tmpdirectory from the
The NTP log files can be filling up the root partition, which is usually omitted. If the NTP services has or had communication issues, the
/var/lib/ntp/var/log/ntpstats directory, which is the common location for NTP logs on all the StorageGRID versions will fill up the root partition. This NTP directory and the
/tmp directory are common reasons for the root partition to fill up. If the NTP services have any communication issue, the
/var/lib/ntp/var/log/ntpstats directory, which is the common location for the NTP logs on all the StorageGRID versions will soon fill up with logs that are about 100 MB or less.
Historically, the reasons above are two major reasons for a root partition (on the StorageGRID) fill up.
Depletion of all the free space on the root partition will prevent proper functioning of the server and might cause file corruption, which can potentially leave the node in an unrecoverable state. To maintain the operating system integrity, it is important to immediately address the space depletion alarms for the root file system.
This article describes the steps to be performed to identify large files in a Linux environment. After identifying these large files, you can determine whether it is safe to move them to an alternate file system (such as the
/var/local/tmp directory), or if they are not required to purge them from the disk.
Although intended for resolving root partition space issues, the principles can be applied to other file systems, such as