The computed limits (limitFactor 3 x request = 1536Mi memory, 30Mi ephemeral
storage) were too small for real documents: a 12 MB HTML already gets the
container OOM-killed, and the caller sees no HTTP error at all, just a dropped
connection and a rising RESTARTS counter.
Rather than raise the requests and reserve conversion-sized capacity on the
node around the clock, requests and limits are decoupled: the requests stay as
they were, and both limits are now set explicitly. The chart already supported
this - an explicit limit wins over the limitFactor computation.
Measured in a throwaway k3d cluster with the official image (HTML via
Chromium), building on the series in the issue:
18 MB HTML, 2Gi -> OOMKilled
18 MB HTML, 2560Mi -> OOMKilled
18 MB HTML, 3Gi -> HTTP 200 in ~22s, 21 MB PDF
12 MB HTML, 1536Mi -> OOMKilled (the old computed limit)
12 MB HTML, 2Gi -> HTTP 200 in ~13s
Memory beyond 3Gi buys nothing: gotenberg's own request timeout is 30s and the
chart passes no flags to change it, so a document needing more memory has
already run out of time (a 25 MB HTML misses the 30s even at 4Gi).
The ephemeral-storage limit is raised in the same change, because raising
memory alone only moves the failure. With 3Gi memory and the old 30Mi limit,
/tmp peaked at 40 MB on a single conversion and at ~125 MB over a short
series, and the pod was evicted:
Pod ephemeral local storage usage exceeds the total limit of containers 30Mi
That failure is worse than the OOM-kill: the pod does not restart, it is
replaced, and the old one is left behind as Evicted.
The defaults in the effectiveResources helper are updated in lockstep with
values.yaml, so an erased "resources:" block (#99) falls back to the new
explicit limits instead of the old computed ones.
NOTES.txt is pulled along. Its numbers came from effectiveResources already,
but the text described a threshold that has moved and advice that no longer
holds: raising the request no longer drags the limit with it now that the
limits are explicit. The warning is kept but no longer claims memory is the
first limit you hit - with these limits, the 30s timeout is.
Major, because the runtime behaviour changes and the pod may occupy more
memory during large conversions than it was allowed to before.
Fixes#106
Retrofit of #99 for a chart that already carries a values.schema.json,
plus the schema type change that goes with it.
An override written without children is YAML null. Where the chart
defines a default, Helm's coalescing deletes the default before schema
validation runs, so the schema never sees the null and cannot reject it:
the container rendered "securityContext: null" - unhardened, healthy,
silent. Where the chart defines no default (resources.limits), the null
survives into validation and a strict "object" type aborted the render.
The defaultedDict helper from #96 handles the first case; widening the
types of exactly these blocks to ["object", "null"] handles the second.
All other keys stay strict, additionalProperties: false is untouched.
The probes address the container port by name, so a rebuilt default probe
needs no further context.
Rendering with normal values is byte-identical to 5.0.0.
Gotenberg has no Ingress, so the notes lead with the cluster-internal URL and
point out that the Service listens on 80 while the container listens on 3000
— the mismatch a consumer such as paperless-ngx has to get right — plus the
port-forward command for reaching it by hand.
Three findings from measuring the running chart went in as warnings:
- The chart passes no command-line flags, so Gotenberg's basic auth, OIDC,
SSRF guards and request timeout cannot be enabled through values at all.
- Memory, not disk, is the first limit to hit: at the default limit a large
document gets the container OOM-killed, the caller sees the connection
dropped rather than an HTTP error, and the notes give the exact kubectl
command that prints OOMKilled.
- Temp space is an emptyDir counted against ephemeral-storage, where running
out looks different — the pod is evicted and replaced, not restarted.
Both limits are read back through the chart's own resources helper, so the
printed figures follow limitFactor and any override instead of being
hardcoded. Refs #43.
jklein