"Your Droplet is nearing its memory capacity" email

[Jonathan5]

I’ve just received the following:

One or more of your Droplets is nearing its memory capacity and its performance could be affected. Without sufficient RAM or memory, applications might run slowly, or can occasionally become unstable and crash.

I’ve a 2GB memory digital ocean droplet.

Over the last 14 days the graph shows about 90-95% memory usage. I asked about memory etc a while back and apparently it’s good to be using what’s available, but this is the first time Digital Ocean has emailed about it. I’m on 2.8.0b6.

The only recent change I can think of is that I set all users to “watch” the main categories, and so receive email notifications, and maybe because of that there have been more posts - but it’s still a relatively quiet forum with about 250 users and a handful of active posters.

What might be causing this, and what should I do, if anything? Or what further information would be helpful? Thanks.

Update. If it helps: disk usage is constant at under 40% and CPU is under 5% with occasional spikes up to about 20%.

[omarfilip]

I received the same email, but none of my droplets are anywhere near capacity. Could be a glitch in DO’s monitoring notifications.

[Jonathan5]

Thank you for the quick response. It could well be a glitch (in which case there might be more replies like yours!) but in my case the graph was bouncing about in the 90-95 range for the full two weeks. I guess the graph could be wrong too…

[omarfilip]

I take that back - I have a test droplet that I hadn’t checked and that’s the one that likely triggered the warning. I’ll rebuild it and see what happens.

[omarfilip]

No significant change after an update, reboot, and rebuild. This is a minimal droplet I use just for testing and it’s normally dormant with CPU at ~1% and no users besides me.

1vCPU / 1 GB Memory / 25 GB Disk / Ubuntu 18.04.3 (LTS) x64

[Jonathan5]

In my case, in the first few minutes at least, it seems to have gone down significantly. I’ll keep an eye on it. I ran ./launcher rebuild mail-receiver followed by ./launcher rebuild app at about 1205.

I wonder what had gone wrong, if anything.

[Jonathan5]

Do you agree that this looks all right now? I’ll still keep an eye on it. The second dip was rebuilding the app again to fix an unrelated problem.

[omarfilip]

It’s not out of the norm. My production instance has 8GB RAM and runs consistently at 41%.

[supermathie (Michael Brown)]

Does DO monitor memory free or memory available?

See: https://www.linuxatemyram.com/

[omarfilip]

DO monitors memory available:

Test droplet

# free -m
         total      used    free    shared  buff/cache available
Mem:       985     687        91        52         206       105
Swap:     2047     236      1811

11% available as shown with free -m and 12% from the monitoring graph, so it’s pretty close (rounding error). This graph is the only measurement the DO panel provides for memory in the Admin panel.

Production droplet

free -m
       total     used     free   shared  buff/cache   available
Mem:    7983     1896     1550     1082        4536        4650
Swap:   4095      291     3804

58% available as shown with free -m and 58% from the monitoring graph.

[j127]

I just started getting those emails on a 2GB (AMD) server too. I rebuilt Discourse a few minutes ago, and memory usage dropped a little.

[j127]

It didn’t last though. At first I thought it might have happened because I recently migrated it from another host to DigitalOcean, but it seemed to be fine for a while after migrating. I’m probably going to upgrade to 4 GB RAM later today.