Draft: CloudHub application sizing and OutOfMemoryError troubleshooting (W-24093783)

[kevintroller]

Summary

• First draft for W-24093783: add an Application Sizing and Memory Management section on cloudhub-architecture.adoc (the ticket’s cloudhub/sizing-guide URL 404s; that page does not exist).
• Draft uses only already published behavior: worker/heap table, 0.1/0.2 vs 1+ vCores for consistent performance, Anypoint Monitoring JVM/Overview charts, automatic restarts, impaired-worker Unrecoverable Runtime Error after five failed starts, streaming/batch OOM causes, and CloudHub 2.0 replica xrefs.
• Do not merge yet. Engineering confirmation is still open in #cloudhub-doc-review. After ENG replies, we can add or drop claims (production-minimum 1 vCore, exit 137/SIGKILL, saw-tooth heap, OOM → Unrecoverable sequence, RTF → CH2 size mapping).

Request vs. draft

Ticket request	Status
vCore memory limits (~500 MB / 1 GB heap)	Already on this page in the CloudHub Workers table — not duplicated; new section points to it
vCores, memory, worker count	Covered; vertical vs horizontal scale with xref to Worker Scale-out
Fractional vCores unsuitable for production / min 1 vCore	Not added — kept existing “1 or more vCores provide performance consistency” pending ENG
Diagnose OOM (exit code 137, SIGKILL)	Not added — unsourced; asked ENG
Saw-tooth heap pattern	Not added — unsourced; asked ENG
Anypoint Monitoring heap / GC / message load	Added UI steps from published Monitoring docs
Streaming / batch for large payloads	Cross-linked Mule runtime topics; not copied
RTF → CloudHub 2.0 sizing	Not added — no official mapping; xref CH2 replica table and CH→CH2 migration only
Exceeding memory → OS kill → Unrecoverable runtime error	Not added as a direct chain — xref impaired-worker (five failed starts) pending ENG

Style

Active voice, contractions (can't, don't), no “the following”. Matched existing CloudHub Architecture heading levels and reused existing xrefs instead of new pages.

Sources

• GUS W-24093783 — case-reduction request (Hive report); target URL 404
• GUS W-24093648 — parent “Create work items from report”
• Published docs: CloudHub Workers table, Impaired Worker Monitoring, Application Monitoring and Automatic Restarts, Anypoint Monitoring built-in dashboards, Mule FATAL_JVM_ERROR / streaming / batch
• Slack #cloudhub-doc-review — ENG confirmation questions (4 Sep 2026)

Test plan

• Local Antora build: new === / ==== headings render under CloudHub Workers
• xrefs resolve: cloudhub-fabric, worker-monitoring, cloudhub-impaired-worker, mule-runtime::mule-error-concept, monitoring::app-dashboards, cloudhub-2::ch2-architecture, ch-ch2-migration-configuration
• Heap-dump partial renders
• After ENG replies: update section for confirmed OOM symptoms only, then SME review