Skip to content

docs(talm): document DRBD sysctl tuning, keepalive toggle, etcd quota #567

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Aleksei Sviridkin (lexfrei) wants to merge 1 commit into
base: main
Choose a base branch
from
+10 −1
Open

docs(talm): document DRBD sysctl tuning, keepalive toggle, etcd quota #567

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 10 additions & 1 deletion content/en/docs/next/install/kubernetes/talm.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -206,7 +206,7 @@ The `cozystack` preset ships curated defaults for `machine.kernel.modules`, `mac
| --- | --- | --- |
| `extraKernelModules` | list | Appended to the built-in modules (`openvswitch`, `drbd`, `zfs`, `spl`, `vfio_pci`, `vfio_iommu_type1`). Each entry is a Talos kernel-module spec. |
| `extraKubeletExtraArgs` | map | Merged into `kubelet.extraConfig` after the preset's `cpuManagerPolicy: static`, `maxPods: 512`. Operator keys must NOT collide with built-ins — yaml.v3 rejects duplicate map keys on decode, so a collision fails the render with a precise hint pointing at the offending key. Fork the preset if you need a different default. |
| `extraSysctls` | map | Merged into `machine.sysctls` after the preset's `gc_thresh*` entries. Same collision-fails-render contract as `extraKubeletExtraArgs`. Values must be YAML strings (Talos expects strings even for numeric sysctls). |
| `extraSysctls` | map | Merged into `machine.sysctls` after the preset's built-in entries: the `gc_thresh1/2/3` ARP-cache sizes, the always-on DRBD/LINSTOR tuning (`tcp_orphan_retries`, `tcp_fin_timeout`, `netdev_max_backlog`, `netdev_budget`, `netdev_budget_usecs`), `vm.nr_hugepages` (when set), and the `tcp_keepalive_*` triplet while `tcpKeepaliveTuning` is enabled. All of these are preset-owned — the same collision-fails-render contract as `extraKubeletExtraArgs` applies. Values must be YAML strings (Talos expects strings even for numeric sysctls). |
Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot Jun 5, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

For clarity and precision, it is recommended to use the fully qualified sysctl names instead of shorthands. This helps operators easily identify the exact keys being configured and avoids any confusion when they configure their own extraSysctls.\n\nFor example, consider using:\n- net.ipv4.tcp_orphan_retries instead of tcp_orphan_retries\n- net.ipv4.tcp_fin_timeout instead of tcp_fin_timeout\n- net.core.netdev_max_backlog instead of netdev_max_backlog\n- net.core.netdev_budget instead of netdev_budget\n- net.core.netdev_budget_usecs instead of netdev_budget_usecs\n- net.ipv4.tcp_keepalive_* instead of tcp_keepalive_*

| `extraMachineFiles` | list | Appended to the preset's CRI customization and `lvm.conf` entries. Talos rejects duplicate `path:` at apply time. |

Example `values.yaml` addition:
Expand All @@ -226,6 +226,15 @@ extraMachineFiles:

The `generic` preset ships no defaults under any of these sections — each block emits only when the matching `extra*` key is non-empty.

Beyond the `extra*` extension points, the `cozystack` preset exposes two opinionated tunables you can change without forking the chart:

| Key | Default | Effect |
| --- | --- | --- |
| `tcpKeepaliveTuning` | `false` | When `true`, adds `net.ipv4.tcp_keepalive_time=600` / `intvl=10` / `probes=6` to `machine.sysctls`, reaping a dead idle socket in ~660s instead of the kernel default ~2h. These sysctls are kernel-wide — they change failure detection for every long-lived idle TCP connection on the node, not just DRBD — so they are opt-in. DRBD already detects dead peers in seconds via its own protocol-level ping, so leave this off unless you specifically want faster node-wide dead-socket detection. |
Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot Jun 5, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

In the description for tcpKeepaliveTuning, the shorthands intvl=10 and probes=6 are used. To make it clear and precise for operators, please use the fully qualified sysctl names:\n- net.ipv4.tcp_keepalive_intvl=10\n- net.ipv4.tcp_keepalive_probes=6

| `etcd.quotaBackendBytes` | `"8589934592"` (8 GiB) | etcd backend DB size ceiling, emitted as `cluster.etcd.extraArgs.quota-backend-bytes` on controlplane nodes only. Raises etcd's own 2 GiB default so a LINSTOR-heavy control plane holding many DRBD-resource CRDs in aggregate does not trip the NOSPACE alarm. It is a ceiling, not a reservation: a small DB stays small and costs no extra RAM/disk. Set it to `""` to fall back to etcd's built-in default. This governs total DB size, not single-object size — per-object writes stay bounded by kube-apiserver's fixed 3 MiB request-body limit, which has no configuration knob. |

The five always-on DRBD/LINSTOR sysctls listed in the `extraSysctls` row above ship unconditionally on the `cozystack` preset — they address TCP-port exhaustion observed under DRBD reconnect storms and have no equivalent on the `generic` preset.

### 2.3 Add Keycloak Configuration

By default, the cluster will be accessible only by authentication with a token.
Expand Down