From bfdc9a64e8bf22e095c4e41a28cc43e0e3487be6 Mon Sep 17 00:00:00 2001 From: Drew Rapenchuk Date: Tue, 1 Aug 2017 06:39:52 -0400 Subject: [PATCH] update scratch.md for issue #4068 (#4069) * update scratch.md for issue #4068 Issue #4068 details the dangers of recommending single etcd instance. This will likely need to be expanded into better "highly available" cluster documentation * Refactor list style * fix ordered list * fix unordered list * ul * add back shell --- docs/getting-started-guides/scratch.md | 35 +++++++++++++++++++------- 1 file changed, 26 insertions(+), 9 deletions(-) diff --git a/docs/getting-started-guides/scratch.md b/docs/getting-started-guides/scratch.md index ffe52a3708..21da67226a 100644 --- a/docs/getting-started-guides/scratch.md +++ b/docs/getting-started-guides/scratch.md @@ -444,14 +444,29 @@ because of how this is used later. - Alternate, manual approach: 1. Set `--configure-cbr0=false` on kubelet and restart. - 1. Create a bridge - - `ip link add name cbr0 type bridge`. + 1. Create a bridge. + + ``` + ip link add name cbr0 type bridge + ``` + 1. Set appropriate MTU. NOTE: the actual value of MTU will depend on your network environment - - `ip link set dev cbr0 mtu 1460` + + ``` + ip link set dev cbr0 mtu 1460 + ``` + 1. Add the node's network to the bridge (docker will go on other side of bridge). - - `ip addr add $NODE_X_BRIDGE_ADDR dev cbr0` + + ``` + ip addr add $NODE_X_BRIDGE_ADDR dev cbr0 + ``` + 1. Turn it on - - `ip link set dev cbr0 up` + + ``` + ip link set dev cbr0 up + ``` If you have turned off Docker's IP masquerading to allow pods to talk to each other, then you may need to do masquerading just for destination IPs outside @@ -501,11 +516,13 @@ all configured and managed *by Kubernetes*: You will need to run one or more instances of etcd. - - Recommended approach: run one etcd instance, with its log written to a directory backed + - Highly available and easy to restore - Run 3 or 5 etcd instances with, their logs written to a directory backed by durable storage (RAID, GCE PD) - - Alternative: run 3 or 5 etcd instances. - - Log can be written to non-durable storage because storage is replicated. - - run a single apiserver which connects to one of the etcd nodes. + - Not highly available, but easy to restore - Run one etcd instance, with its log written to a directory backed + by durable storage (RAID, GCE PD) + **Note:** May result in operations outages in case of instance outage + - Highly available - Run 3 or 5 etcd instances with non durable storage. + **Note:** Log can be written to non-durable storage because storage is replicated. See [cluster-troubleshooting](/docs/admin/cluster-troubleshooting) for more discussion on factors affecting cluster availability.