2021 Update Rollup 2

Scaling with Split PointsPermanent link for this heading

To enable a distribution of a CAS area, so-called split points can be created. The distribution of data is processed depending on the calculated hash values.

Example for split point configuration:

To distribute a CAS area to three different storage devices, three split points can be created directly in the defined Fabasoft Folio MMC Area. A split point can be a directory or a file. If the split point is a directory then this directory will be used for saving the corresponding data. In case of a split point file the path specified in this file will be used. Such a file can contain a network path or a local directory path. A split point has the following syntax: splitpoint_<xx>. <xx> represents the first two characters of the hash value.

In this example the split points are “00”, “55” and “aa”. By means of this structure all hash values starting between “00” and “54” are stored in the directory splitpoint_00, all hash values between “55” and “a9” in the directory splitpoint_55 and all hash values from “aa” in the directory specified in the splitpoint_aa file. According to this illustration it is possible to create up to 256 split points.

If the splitpoint_00 split point is not created the content is located directly in the defined Fabasoft Folio MMC Area directory.

Split points can be created or deleted as necessary. The right distribution of the content (changes in the file system have to be processed manually) and a restart of Fabasoft Folio MMC Services (the split point structure is only read at first access) are important for successful modification.

Note: The consistence of split points is checked after first access to a CAS area. In cases of inconsistence (files are in the wrong split point directory) an error message is written to the event log.

In a Linux system environment, so-called “symbolic links” can be used to locate split point directories on different storage devices. In a Microsoft Windows system environment, a so-called “mount point” can be created on these directories.

If file shares are used as MMC or split point directories, the system administrator must ensure that these directories are mounted before the MMC service or the Kernel accesses the CAS area for the first time. In Linux the configuration file /etc/fsctab can be used to specify all mounts. If one mount requires another, the option x-systemd.requires-mounts-for can be used in the fstab config file.

Example fstab entries:

//10.10.10.1/mmc /data/mmc cifs domain=mydomain,username=user,x-systemd.requires-mounts-for=/data/splitpoints 0 2
//10.10.10.2/splitpoints /data/splitpoints cifs domain=mydomain,username=user 0 2

Furthermore, it is possible to define dependencies for service units to ensure required mountpoints are available before starting a service. To add a specific mountpoint as dependency for the fsc.service an override.conf file needs to be created in the fsc.service.d folder. This can be done by executing the following command:

systemctl edit fsc.service

For every required mountpoint the option After= and Required= with the associated mount file needs to be added in the [Unit] section.

Example override.conf file:

[Unit]
After=data-mmc.mount
Requires=data-mmc.mount

The mount files (units) are automatically generated by systemd for every fstab entry. To get the unit names of all mountpoints, execute the following command:

systemctl list-units --type=mount

Reload the systemd daemon to make the changes take effect.

systemctl daemon-reload