diff --git a/10-systems/hosts/nextcloud.md b/10-systems/hosts/nextcloud.md new file mode 100644 index 0000000..1396975 --- /dev/null +++ b/10-systems/hosts/nextcloud.md @@ -0,0 +1,34 @@ +--- +owner: gmikcon +status: active +last_reviewed: 2026-10-07 +review_interval: 90d +confidence: high +source_of_truth: ssh-host-verified +--- + +# Nextcloud VM + +Nextcloud file cloud VM, upstream for `cloud.gnexus.space`. + +## Identity + +- Inventory id: `nextcloud`. +- VM id: `nextcloud` (libvirt domain on hypervisor `hp-proliant-dl380-g6`, bserv `192.168.1.130`). +- LAN address: `192.168.1.152`. +- Resources: 24 vCPU, 32 GiB RAM. +- Disk: qcow2 `/mnt/main/vm_main/nextcloud`. +- Autostart: enabled. Note: libvirt autostart only fires on host/libvirt start — it does NOT recover a VM killed by the OOM killer. + +## Roles + +- Nextcloud instance behind `cloud.gnexus.space` (TLS terminates on `internal-proxy-vps`, upstream `http://192.168.1.152`). + +## Incident 2026-10-06 (OOM crash) + +- 2026-10-06 07:53:20 UTC: global OOM on hypervisor bserv — 117 GiB RAM exhausted and 8 GiB swap full. Kernel OOM killer killed the nextcloud qemu process (pid 3965, ~28 GiB RSS, largest allocation on the host). +- 2026-10-06 07:54:06 UTC: VM shut down, `reason=crashed` (libvirt log). +- Down ~31h12m; `cloud.gnexus.space` returned 502 from the internal proxy during the outage. Started manually by the owner 2026-10-07 15:06 UTC; service verified 200 after start. +- Standing risk: hypervisor memory stays tight (90/117 GiB used, swap 8/8 GiB full at check 2026-10-07); the nextcloud VM is the largest single memory allocation (32 GiB) and the most likely OOM victim again. + +No raw credentials are documented here. diff --git a/90-maintenance/pending-changes/20261007150805-1da4beb7.json b/90-maintenance/pending-changes/20261007150805-1da4beb7.json new file mode 100644 index 0000000..2ee17c5 --- /dev/null +++ b/90-maintenance/pending-changes/20261007150805-1da4beb7.json @@ -0,0 +1,19 @@ +{ + "id": "20261007150805-1da4beb7", + "kind": "doc", + "target": "10-systems/hosts/nextcloud.md", + "summary": "Add nextcloud VM host doc: identity, cloud.gnexus.space upstream role, 2026-10-06 OOM crash incident.", + "reason": "Host 192.168.1.152 (cloud.gnexus.space upstream) was missing from inventory; OOM crash 2026-10-06 diagnosed from libvirt/kernel logs on bserv.", + "payload": { + "content": "---\nowner: gmikcon\nstatus: active\nlast_reviewed: 2026-10-07\nreview_interval: 90d\nconfidence: high\nsource_of_truth: ssh-host-verified\n---\n\n# Nextcloud VM\n\nNextcloud file cloud VM, upstream for `cloud.gnexus.space`.\n\n## Identity\n\n- Inventory id: `nextcloud`.\n- VM id: `nextcloud` (libvirt domain on hypervisor `hp-proliant-dl380-g6`, bserv `192.168.1.130`).\n- LAN address: `192.168.1.152`.\n- Resources: 24 vCPU, 32 GiB RAM.\n- Disk: qcow2 `/mnt/main/vm_main/nextcloud`.\n- Autostart: enabled. Note: libvirt autostart only fires on host/libvirt start — it does NOT recover a VM killed by the OOM killer.\n\n## Roles\n\n- Nextcloud instance behind `cloud.gnexus.space` (TLS terminates on `internal-proxy-vps`, upstream `http://192.168.1.152`).\n\n## Incident 2026-10-06 (OOM crash)\n\n- 2026-10-06 07:53:20 UTC: global OOM on hypervisor bserv — 117 GiB RAM exhausted and 8 GiB swap full. Kernel OOM killer killed the nextcloud qemu process (pid 3965, ~28 GiB RSS, largest allocation on the host).\n- 2026-10-06 07:54:06 UTC: VM shut down, `reason=crashed` (libvirt log).\n- Down ~31h12m; `cloud.gnexus.space` returned 502 from the internal proxy during the outage. Started manually by the owner 2026-10-07 15:06 UTC; service verified 200 after start.\n- Standing risk: hypervisor memory stays tight (90/117 GiB used, swap 8/8 GiB full at check 2026-10-07); the nextcloud VM is the largest single memory allocation (32 GiB) and the most likely OOM victim again.\n\nNo raw credentials are documented here." + }, + "status": "applied", + "created_at": "2026-10-07T15:08:05Z", + "updated_at": "2026-10-07T15:08:07Z", + "applied_at": "2026-10-07T15:08:07Z", + "validation": { + "status": "ok", + "issue_count": 0, + "issues": [] + } +}