I do development for a charity that sends re-purposed machines configured to behave as wireless access points and loaded with kolibri and other learning resources to underprivileged countries to empower classrooms to educate their students.
this is built on a Linux system and the data is being stored on an SD_CARD
KOLIBRI_HOME is pointed to a directory on this SD_CARD
my question is, does kolibri store user and facility data, in the KOLIBRI_HOME directory as well. that is user accounts, progress, grades, all the actual DATA that is generated while using the system. is this stored in the KOLIBRI_HOME directory?
We are tight with the amount of free space on the SD Cards and need to know if we need to leave space in the partition for this type of data for systems once deployed.
IF this data is stored in KOLIBRI_HOME, might someone be able to give me an idea of HOW MUCH space that data might need. both considering classroom sizes, and over multiple semesters…I don’t want the systems to seize in the field because of failure to be able to write to that partition.
and finally, I figure I should ask, is there a daemon variable like KOLIBRI_HOME to specify a different storage location for this user generated data?
We are currently using Kolibri Version: 0.17.5, running on Debian Bookworm, and deployed on Dell Chromebook 3189 (Kefka)
Thank You for your responses and assistance in this matter!
Welcome to the Learning Equality forums, and for your interest in Kolibri!
Yes, by default Kolibri stores all of its data in the KOLIBRI_HOME directory.
The user and application data is generally not of significant size. Overall, that data is dwarfed by the disk usage for content imported into Kolibri. Unless you’re planning for these devices to serve many thousands of users, it’s unlikely that the Kolibri application data would grow over a few hundred megabytes. If you haven’t already, I would suggest taking a look at this forum category to see if others have mentioned this. You can of course post this particular question there too, but I welcome others to share their insights here too.
The KOLIBRI_HOME directory can be specified by exporting an environment variable with the same name. The Kolibri documentation has information on how to achieve that. Although, you may be more interested in specifying where imported content is stored instead, which is controlled by the CONTENT_DIR option in Kolibri’s options.ini.
Lastly, Kolibri has a Debian installer and a kolibri-server package which both manage the KOLIBRI_HOME directory in a manner that is explicitly controlled and setup at time of installation.
I do apologize for my delayed response. I have a sleep disorder that causes me to sleep excessive periods of time…
The information you provided was EXACTLY what I needed. I thank you for your reply. I do have ONE more question of you. the few hundred megs, would that include any user generated data from large channel platforms like Khan Academy? so would it be a few hundred megs even with heavy use of large systems like that? or would a channel package like khan academy add exponentially more of this user generated data?
I don’t mean to be a pain. just want to know what to expect is all. I thank you VERY much for your time and response!
When a channel is imported into Kolibri, its metadata is added to the primary Kolibri database. This means the amount of imported content does contribute to the database size, but that should be a fixed cost at the time of import. One of the largest Khan Academy channel databases is around 120MB, so you can expect that amount to be contributed to the primary Kolibri database file when it’s imported. Although, that size is still dwarfed by the channel’s content files, which are around 100GB in total for that same channel.
I dug up an old Kolibri database file. The file size is around 320MB. The Kolibri Facilities on it have 170 users total, around 150% more imported content than that of the aforementioned channel, and the content has around 30k total interactions. Overall, the imported channel metadata represents 75% of the database file size. So while the ongoing growth of the database is primarily tied to users utilizing Kolibri, it still isn’t the most significant contribution to disk usage overall. Nevertheless, the quantity of users and the time scale of your implementation will matter most in terms of how much space you’ll need for growth.
If you expect thousands of users per Kolibri instance, performance may start to suffer with the default database backend of SQLite. It is possible to configure Kolibri to use PostgreSQL, but the disk usage in that approach would no longer be contained to KOLIBRI_HOME.