I restarted the service and things appear to be working atm.. but I don't understand how the volumes are configured on the machine so will have to wait for Ben to fix it properly. On Mon, Oct 18, 2021 at 9:11 AM Zubin Duggal wrote:

It is not just you, Gitlab has been unstable since yesterday due to a lack of disk space.

On 21/10/18 07:29, Simon Peyton Jones via ghc-devs wrote:

...

I'm getting "502" from gitlab.haskell.org. Is it just me?

Thanks

Simon

PS: I am leaving Microsoft at the end of November 2021, at which point simonpj@microsoft.commailto:simonpj@microsoft.com will cease to work. Use simon.peytonjones@gmail.commailto:simon.peytonjones@gmail.com instead. (For now, it just forwards to simonpj@microsoft.com.)

...

_______________________________________________ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

_______________________________________________ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

Tom Ellis writes:

If a sponsor (perhaps the HF) could pay for GitLab to host the service on our behalf would that be helpful? I don't know whether GHC development relies deeply on some aspect of our self-hosted setup. (I suspect it does because otherwise we'd likely be using a free GitLab tier for open source organisations, but I thought the question was worth asking.)

Migrating towards the SaaS offering is indeed quite enticing. However, last time I checked it wasn't clear to me that it was feasible in our case. The problem isn't the cost as we could in principle request Ultimate-level access to the SaaS via GitLab's FOSS program (which is the same program through which we have our current self-hosted license). The original reason was that we needed to patch GitLab during the migration away from Trac to ensure that ticket metadata was faithfully preserved. While at this point we are running a completely unmodified GitLab instance, my understanding is that there are a few sticking points that would make migration to the SaaS difficult: * the default 10 GB storage quota is extremely small when one considers the size of our CI logs, artifacts, and Docker images * some administrative and debugging tasks sadly become difficult without server access (e.g. I've sometimes found it helpful to look at the server log while debugging CI issues) * Response time of GitLab.com's search functionality has historically been extremely variable, rendering it at times unusable. I'm not sure whether this has improved recently. The first point is especially concerning and has been the primary source of our troubles in self-hosting. As long as we don't run out of storage, GitLab administration tends to be a very minor time commitment (an hour or so a month dealing with spam and upgrading). Thankfully, after today I think we have sufficiently over-provisioned to avoid any future storage issues for the foreseeable future. Cheers, - Ben