GitLab.com melts down after wrong directory deleted, backups fail

So in other words, out of 5 backup/replication techniques deployed none are working reliably or set up in the first place.

Oh dang! Events like these highlight the value of self hosting.

But I can only imagine the stress levels over there right now. :frowning:

Are you sure? How many would-be self-hosters are more competent than a *aaS company at backups and all other forms of resilience (e.g., against private and government criminals)?

Backup for whatever your domain of concern is and have a plan to failover or migrate, and test those facilities, for sure.

Ah, yes I was talking about just me. I am competent enough to make regular backups of what I care about, and verify their usability. In that sense, yes I am sure. :slight_smile:

But I see your point.

Yep! Self-hosters are able to scale at their own needs, whereas gitlab.com is a free service used as a promotional tool and technical demo, and thus has way more scale than any single company would be expected to handle. Our little code site is easily managed by me (and I check that backups are made, but I don’t really test them, because for my use I already have a copy of all the important info in my repo locally…)

Also, making more targets is one method for combating bad actors (both private and government). Granted that scripting can take out a bunch of sites, so that is a case by case basis.

What’s it called… what is that premise that multiple targets make it harder to attack? I feel like there is a specific term for it, but I can’t find it with my search. Anyhow, self-hosting in particular, and GitLab specifically, benefit from that. Like GitHub before it, having a single point of failure for a large amount of code isn’t the best plan.

I ran into trouble with the first incarnation of ATC, because I was partially-hosting it through a company that provided GitLab hosting. But they didn’t last, and self-hosting GitLab is actually a fairly straight-forward process, if one is familiar with ssh.

Aside: remember Gitorious? That was fun. And technically the entity we are discussing.

It is an interesting topic, to be sure. I wouldn’t say self-hosting a federated social network is a good idea, despite my own personal history is self-hosting all of them. But generally I think self-hosting is more resilient than a single *aaS provider, despite their relative competency. :slight_smile:

1 Like

Well, having a single point of failure for a small amount of code is what self-hosting is, I mean without other copies anyway.

In either case, only lots of copies in lots of places (ie not single instance, whether hosted by an individual or a giant company) is the only way to keep stuff safe.

One of the sad things is that centralized things get noticed when they go belly up, and the Archive Team tries to save them (IIRC they played a role in saving gitorious data). Self-hosted stuff just falls off the net at a rapid clip and doesn’t get saved other than by the Wayback Machine, which is great but doesn’t help for application data. I’d love to see archiving (for posterity, not backup restore) more available to self-hosted/small sites to correct this imbalance. For software, https://www.softwareheritage.org/ is an interesting project, akin to Internet Archive for software.

A few other things about self-hosting as a general solution: it’s out of reach for vast majority of people who just lack the knowledge, time, resources, social networks to gain these, and it is an investment even for people who have all of those that one gets sick of making, eg a few days ago I saw One Small Voice: VPS No More (someone with decades of experience self-hosting giving it up to simplify). And yet another problem is that even a menacing faceless corporation can be better than relying on technical family or friends to help one self-host – family or friends who may be abusive – much like moving to the city means dealing with faceless and possibly crushing institutions, but at least the villagers aren’t individually spying on you and worse.

My bottomline: more commons-oriented institutions are needed (http://platformcoop.net/ is interesting take on this); self-hosting is great for some people but as a general response fails as it is atomistic.

Lots to read, but I wanted to share a thought I’ve had recently, and will digest the rest of yer message later.

Actually, it may be what “Platform Cooperativism” is about…

Anyhow, I don’t self-host for myself, and I don’t expect others to do that either. I expect folks to group together. With a very shallow reading of historically autonomous family units, not everyone was only a farmer; folks each have their own crafts and expertise.

So when I talk about self-hosting I speak about it from the perspective of both a person that does knowledge work and hosting professionally, but also as a designated village expert for hosting needs.

I am not great at communicating that, but I am getting there. Since I’ve gone back to consulting full-time, I’ve had a lot of discussions (some of them recorded) so I can better understand how to have a positive impact for the people around me. And I know other folks are working in the same realm (like the aforementioned platform coop). :slight_smile:

1 Like

FWIW Platform Cooperativism is about lots more than web stuff, but interesting to think about such as analogy. One of the classic subjects is transportation. Car-sharing is an interesting service that could be controlled by Uber, or could be run by cooperatives (with free software, preferably). What’s the self-hosting of car-sharing? Ad-hoc carpooling? Jitneys? These seem not up to taking on Uber, neither politically nor actually providing what good the service can provide.