mirror of
https://github.com/rubenhensen/k8scd.git
synced 2026-11-16 06:29:09 +01:00
docs: WordPress is capped at 12 workers
This commit is contained in:
1 parent
7dc13466cb
commit
714712b862
1 file changed
+14
-8
+14
-8
@@ -280,16 +280,22 @@ Counted 2026-10-25 over the 73 running containers:
|
|||||||
kube-system
|
kube-system
|
||||||
- [ ] Memory limits, once there are daytime numbers: the night ones miss
|
- [ ] Memory limits, once there are daytime numbers: the night ones miss
|
||||||
immich's machine learning at work (it idles at 234Mi)
|
immich's machine learning at work (it idles at 234Mi)
|
||||||
- [ ] WordPress (scoutingmbg.nl) was OOMKilled at 08:27:59 UTC on
|
- [x] WordPress (scoutingmbg.nl) was OOMKilled at 08:27:59 UTC on
|
||||||
2026-10-25, its third restart: a crawler asked for 88 pages and assets
|
2026-10-25, its third restart: a crawler asked for 88 pages and assets
|
||||||
within a few seconds (normally ~6 a minute), Apache started a PHP worker
|
within a few seconds (normally ~6 a minute), Apache started a PHP worker
|
||||||
for each, and the container went past its 1Gi limit (500s first, then
|
for each (~45-60 MB each, `MaxRequestWorkers` was the default 150), and
|
||||||
the kill; back within seconds). Unrelated to the NetworkPolicies; memory
|
the container went past its 1Gi limit. Unrelated to the
|
||||||
sat at ~500Mi before. Fixes, in order: cap Apache's prefork workers
|
NetworkPolicies. Now capped at 12 workers through the chart's
|
||||||
(`MaxRequestWorkers` ~16, so a burst queues instead of filling memory),
|
`apachePortsConfig` (ports.conf comes after mods-enabled/*.conf), checked
|
||||||
a page cache plugin. Rate limiting per visitor in Envoy doesn't work
|
with `apache2ctl -t` in the same image first. Replaying the burst (90
|
||||||
yet: node001 passes TLS through, so every request arrives from
|
page requests, 40 at a time): all 200, at most 13 Apache processes and
|
||||||
46.224.26.65; that needs the PROXY protocol from node001's nginx to Envoy
|
572Mi, no restart; the price is waiting (4.3s median, up to ~22s at the
|
||||||
|
peak) instead of a crash
|
||||||
|
- [ ] A page cache plugin for WordPress: a page takes seconds to render,
|
||||||
|
which is what makes a burst queue up
|
||||||
|
- [ ] Rate limiting per visitor in Envoy needs the client's address, which
|
||||||
|
node001 loses: it passes TLS through, so every request arrives from
|
||||||
|
46.224.26.65. That needs the PROXY protocol from node001's nginx to Envoy
|
||||||
(ClientTrafficPolicy `enableProxyProtocol`)
|
(ClientTrafficPolicy `enableProxyProtocol`)
|
||||||
- [x] Probes for gitea (`/api/healthz`), freshrss (TCP, since
|
- [x] Probes for gitea (`/api/healthz`), freshrss (TCP, since
|
||||||
mod_auth_openidc answers 401 to non-browsers) and serpbear (`/login`)
|
mod_auth_openidc answers 401 to non-browsers) and serpbear (`/login`)
|
||||||
|
|||||||
Reference in new issue
Block a user