Skip to content
Register

systemd service keeps dying on 256MB RAM Alpine container

VPS Hosting by tamuma 14 replies 1.8K views
Marek

Member

Traceroute Detective

6 #11
tamuma said:
journalctl shows "Failed to allocate memory" then freeze

Can you strace the service start? I want to see if it fails on mmap or on clone. Also check /proc/user_beancounters if OpenVZ, failcnt column tells real story. Provider hides limits there.

I saw same on my Wrocław lab box, turned out privvmpages limit was 280MB but kmemsize tiny. Systemd allocates lot of kernel memory for cgroups.

JokoNord

Member

#12
tamuma said:
Provider is RamNode, not scam

Ah RamNode. I have client on RamNode OpenVZ 512MB, same silent death. Their beancounters kmemsize very low. OpenVZ 6 ancient kernel. You cannot win with systemd here.

Switch to OpenRC or move to KVM. RamNode KVM 512MB is $5, still cheap. Or Time4VPS 512MB €3.99 as DmitriEng say.

traffic worse than my packet loss
RO

Member

1 #13

Following this. Have same setup on VirMach and same problem.

FL

Member

1 #14
lucgone said:
OpenRC uses 3MB. S6 maybe 1MB. Why fight it?

This. I run s6 on 256MB Alpine at Mythic Beasts in London. Zero issues. Systemd is desktop software, fight me.

Also RamNode OpenVZ is dead platform. They keep it for old customers. KVM or leave.

JA

Member

9 #15
JokoNord said:
Check memory.limit_in_bytes

Already covered, it's 256MB. But beancounters is right place to look.

Anyone tried running systemd with systemd.unified_cgroup_hierarchy=0 on Alpine? I heard legacy mode uses less memory. Not tested myself.

Post a reply

You need an account to reply. Log in or register to join the conversation.

Post reply Preview Save draft