cal/claude-memory
1
0
Fork 0
Code Issues 2 Pull Requests Actions Packages Projects Releases Wiki Activity

Compare commits

base: cal:955f0206c5465c429f2dad3a0ccb7cd3e555b325
Branches Tags
cal:main
...
compare: cal:f2c6cd4bf80ffa9f45a8b2ac75241f499c47e30a
Branches Tags
cal:main

2 Commits
955f0206c5 ... f2c6cd4bf8

Author SHA1 Message Date
Cal Corum
f2c6cd4bf8 episode: LXC load_multiplier threshold too low - bumped from 2 to 4 on CT 302 2026-02-20 08:53:54 -06:00
Cal Corum
a2358bd123 store: LXC load_multiplier threshold too low - bumped from 2 to 4 on CT 302 2026-02-20 08:53:54 -06:00
2 changed files with 38 additions and 0 deletions
Show Stats Expand all files Collapse all files

4
episodes/2026-02-20.md
View File

@ -69,3 +69,7 @@ Session size: 10 messages, 2 tool calls
- **Memory:** [homelab-session-automation-configuration-creation-deployment-197640](graph/fixes/homelab-session-automation-configuration-creation-deployment-197640.md)
- **Summary:** Work types: automation, configuration, creation, deployment, fix
Session size: 74 messages, 17 tool
## 08:53 - LXC load_multiplier threshold too low - bumped from 2 to 4 on CT 302
- **Type:** fix
- **Tags:** monitoring, claude-runner-monitoring, proxmox, lxc, configuration, homelab

34
graph/fixes/lxc-load-multiplier-threshold-too-low-bumped-from-2-to-4-on-4da1b6.md Normal file
View File

@ -0,0 +1,34 @@
---
id: 4da1b677-8bb6-445c-afc9-7186e5898f29
type: fix
title: "LXC load_multiplier threshold too low - bumped from 2 to 4 on CT 302"
tags: [monitoring, claude-runner-monitoring, proxmox, lxc, configuration, homelab]
importance: 0.6
confidence: 0.8
created: "2026-02-20T14:53:54.422260+00:00"
updated: "2026-02-20T14:53:54.422260+00:00"
---
# LXC load_multiplier Threshold Fix on CT 302
## Problem
`health_check.py` on CT 302 (claude-runner, 10.10.0.148) was generating constant "load exceeds" warnings for all 3 monitored servers: arr-stack, gitea, and uptime-kuma.
## Root Cause
All monitored servers are 2-core LXC containers. LXCs share the host kernel's `/proc/loadavg`, so the load numbers reported inside a container reflect the **entire Proxmox host**, not just that container. With `load_multiplier=2`, the threshold was `cores×2=4`, which was consistently exceeded by normal host activity (loads of 3-5 across all containers).
## Fix
Bumped `load_multiplier` from `2` to `4` in `/root/.claude/skills/server-diagnostics/config.yaml` on CT 302:
```yaml
load_multiplier: 4 # was 2
```
New threshold: `cores×4=8`, which only fires on genuinely concerning spikes.
## Key Insight
LXC containers report host-wide load averages via `/proc/loadavg`. Per-container core-based thresholds must account for this by using a higher multiplier (4x or greater) to avoid constant false positives from normal host activity.