I'm looking to setup a light SIEM in my homelab, well it's more just an "observability" setup really. I'm planning to put both logs and metrics into this across a variety of systems. I know I can export metrics via the native influxdb/graphite options but there's no built-in way to export logs.
I'm going to use Grafana Alloy | Grafana Alloy documentation to collect system logs, but I know the advice is to put a minimal amount of stuff on top of proxmox to prevent any issues with updating or dependency conflicts. My idea was that I would spin up an lxc on each pve node and bind mount proxmox's `/var/logs/` directory into that lxc, hopefully as read-only, and then install the Alloy agent inside the lxc.
Would I run into any issues doing that? I think there might be, hopefully minor, permission conflicts as it should be read-only to the LXC, so alloy isn't going to be writing to it on either the PVE side or the LXC side.
If I can get away with running the Alloy agent 'bare metal' on the PVE nodes directly, then that could be useful as Alloy also collects system metrics, as well as traces, profiling and "frontend user monitoring". I'm not sure how useful those last 3 would actually be for proxmox monitoring, but it could be cool to play with.
If I do go the LXC route for Alloy then I won't be able to collect system metrics, but I can probably just use the promethis-pve-exporter project for those (if I don't fall back on influxdb, which is an option).
I'm going to use Grafana Alloy | Grafana Alloy documentation to collect system logs, but I know the advice is to put a minimal amount of stuff on top of proxmox to prevent any issues with updating or dependency conflicts. My idea was that I would spin up an lxc on each pve node and bind mount proxmox's `/var/logs/` directory into that lxc, hopefully as read-only, and then install the Alloy agent inside the lxc.
Would I run into any issues doing that? I think there might be, hopefully minor, permission conflicts as it should be read-only to the LXC, so alloy isn't going to be writing to it on either the PVE side or the LXC side.
If I can get away with running the Alloy agent 'bare metal' on the PVE nodes directly, then that could be useful as Alloy also collects system metrics, as well as traces, profiling and "frontend user monitoring". I'm not sure how useful those last 3 would actually be for proxmox monitoring, but it could be cool to play with.
If I do go the LXC route for Alloy then I won't be able to collect system metrics, but I can probably just use the promethis-pve-exporter project for those (if I don't fall back on influxdb, which is an option).