Astiga downtime 2021-03-30

The Astiga user count has increased immensely recently, thus putting more pressure on the server. I was in the process of making a plan how to tackle that issue (which was planned for the end of April), and then the server decided that it had enough. This is the server load of the server on 2021-03-30 (and 2021-03-31).

The server load is essentially the amount of processes waiting to get CPU time, generally expressed as an average in 1 minute, 5 minutes, and 15 minutes. As a rule of thumb, you want the load to be less than the amount of CPU cores you have, as having a load that is higher than that means that the amount of processes awaiting processing grows and grows. Normally it should be around 2.

Astiga runs on two servers: one server that just does synchronisation, and one server that does everything else. At the start of today, the synchronisation server ran on a 4 core, 8 GB RAM machine. The "everything else" server ran on a 2 core, 4 GB RAM machine. Yes, that is probably slower than your laptop or computer, but then everything in Astiga has been optimised to work with little processing power, to keep the costs low for you and for me. As of now, both run on a 4 core, 8 GB RAM machine. Hopefully doubling the performance suffices.

I am not fully sure what the direct cause was of the load increase - it might have been as simple as a process syncing the podcasts, or someone loading a lot of album art.

Astiga was down for a bit over an hour, which is more than I'd have liked. Most time was spent on creating a snapshot of the server prior to upgrading it (it runs at DigitalOcean, if you are interested), just to have a way to revert in case the upgrade went wrong.

My apologies for the unexpected downtime. Should your sync have been running around that time, then it will automatically restart in a couple of hours. If not, you should be able to restart it yourself without having to wait.

The planned maintainance will most likely still take place in April, but more about that later.