Join Log in
Score:1
WilhelmM avatar Server

Using git lfs to manage data in storage server

cn flag
WilhelmM

I am part of a small physics research team (10-15 people) which recently has acquired a storage server and I will be responsible for setting it up. As such, I needed to think the best way to organize the data we will be generating.

When researching the best way to do this, I came across git lfs. We already use GitLab (the cloud version) to version control our codes and easily share them between us. So, it would make sense to create a "data" repository where most of the files will be binary ones, stored and tracked via lfs. However, this seems that it will upload the files to GitLab's server, defeating the whole purpose of a storage computer (also, we would need to buy storage from them).

The solution seems to be for us to deploy the community edition of GitLab on our storage server. Now comes my questions:

  1. Is that the only way? Ideally, I would like to be able to tell GitLab that our files are stored elsewhere, and the git repo just point out that "elsewhere" is our storage server?
  2. If I must deploy GitLab's CE, how hard is to maintain such a server? My ideal scenario is a zero maintenance, i.e. setup once and, as long no one updates the OS, it should not break (see comment below).
  3. In the case it breaks, will the data be stored in a way someone can easily retrieve, i.e. it will not be compacted in a binary blob where you need the server running to retrieve it. Ideally, I should be able to even read the data from within the server with the service running, since we may do some data analysis inside this computer.

Comment about point 2: I know not performing regular updates may leave security holes. However, the data stored does not contain sensitive information. We do not deal with personal information and leaks or data loss would have only minor consequences. On the other hand, I am a post-doc and may need to leave the group at any moment in favor of a permanent position. Since the group is focused on physics research, other people that come after I leave may not have the necessary skills for maintaining the server. Thus, the need for a system to manage the data that requires little to no maintenance.

15
0 + 0
Score:0
avatar Server
jp flag
Sekenre

You have two options in my mind.

  1. Run self-hosted community Gitlab on your storage server, and use git-lfs on that.
  2. Use git-annex.

Git-annex tracks the file locations in git but the actual file content can be stored in a wide variety of backends, eg; rsync, s3, bup. One of which you could set up on your storage server: https://git-annex.branchable.com/special_remotes/

Edit to answer your question about Gitlab CE Admin. We use it for a small dev team of 10 people, and around 100 repos. If you are not using Gitlab's Continuous Integration, admin should be very simple.

0 + 0
ธงชาติไทย
Elon Musk
I sit in a Tesla and translated this thread with Ai:

mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.