Bonjour,
For the record I increased the number of vCPUs on the GitLab runner, from 1 to 4. It should speed up things during the creation of the GItLab instances. There is enough memory and the disk is fast, the bottleneck was the CPU.
Cheers
There may not be enough memory after all. The following happened today:
E STDERR:
E remote: fatal error: runtime: out of memory
E remote:
E remote: runtime stack:
E remote: runtime.throw(0xce8313, 0x16)
E remote: /usr/local/go/src/runtime/panic.go:1116 +0x72 fp=0x7fffa2d60160 sp=0x7fffa2d60130 pc=0x438332
E remote: runtime.sysMap(0xc000000000, 0x4000000, 0x12bb6d8)
E remote: /usr/local/go/src/runtime/mem_linux.go:169 +0xc6 fp=0x7fffa2d601a0 sp=0x7fffa2d60160 pc=0x41aea6
E remote: runtime.(*mheap).sysAlloc(0x12a0220, 0x400000, 0x0, 0x0)
E remote: /usr/local/go/src/runtime/malloc.go:727 +0x1e5 fp=0x7fffa2d60248 sp=0x7fffa2d601a0 pc=0x40e6c5
E remote: runtime.(*mheap).grow(0x12a0220, 0x1, 0x0)
E remote: /usr/local/go/src/runtime/mheap.go:1344 +0x85 fp=0x7fffa2d602b0 sp=0x7fffa2... (2822 more, please see e.stderr)
.tox/py3/.venv/lib/python3.8/site-packages/sh.py:905: ErrorReturnCode_1
I’m unsure of your infra sponsor, however as mentioned on today’s call we (Gitea) have Equinix metal as a sponsor and you can also apply via https://metal.equinix.com/solutions/oss/ in case you require additional resources (they also sponsor many projects such as golang, nix, wireguard, and all CNCF projects such as kubernetes).
More info here: https://metal.equinix.com/solutions/oss/
2 Likes
It’s good to know, thanks! The current hardware is donated by Easter-Eggs, a French Cooperative dedicated to Free Software.