Same experimental setup as https://gist.github.com/cgwalters/64752a9544eec662e056f8387a3cefc2#file-ssd-test-md
Based on
[global]
bs=2000
ioengine=sync
iodepth=1
runtime=60
directory=/mnt
From ffc795562602637a7c3bc57bad7f20fdad58fa33 Mon Sep 17 00:00:00 2001 | |
From: Stephen Lowrie <[email protected]> | |
Date: Thu, 3 Oct 2019 15:40:48 -0500 | |
Subject: [PATCH] HACK: Manhole on failure | |
--- | |
kola/cluster/cluster.go | 4 ++++ | |
1 file changed, 4 insertions(+) | |
diff --git a/kola/cluster/cluster.go b/kola/cluster/cluster.go |
#!/bin/bash | |
# forked from https://gist.github.com/innovia/fbba8259042f71db98ea8d4ad19bd708 | |
set -e | |
set -o pipefail | |
# Add user to k8s using service account, no RBAC (must create RBAC after this script) | |
if [[ -z "$1" ]] || [[ -z "$2" ]]; then | |
echo "usage: $0 <service_account_name> <namespace>" | |
exit 1 | |
fi |
``` | |
walters@toolbox ~/s/g/c/coreos-assembler> git diff | |
diff --git a/src/create_disk.sh b/src/create_disk.sh | |
index e9cd8596..76808b3e 100755 | |
--- a/src/create_disk.sh | |
+++ b/src/create_disk.sh | |
@@ -10,6 +10,9 @@ | |
# an issue and we can discuss configuration needs. | |
set -euo pipefail | |
#!/bin/bash | |
# Automate an end-to-end run of coreos-installer with the metal image, which then | |
# boots and writes a success message to a virtio-serial port, which we read on the host. | |
set -euo pipefail | |
run_tmp_webserver() { | |
local statedir=$1 | |
shift | |
local dir=$1 | |
shift |
On host in your cosa build directory run e.g.: | |
``` | |
$ cd /srv/walters/rhcos-4.4 | |
$ python3 -m http.server | |
``` | |
``` | |
$ cosa run -d rhcos-4.2.0-qemu.qcow2 | |
# ostree remote add --set=gpg-verify=false local http://10.0.2.2:8000/tmp/repo | |
# while ! rpm-ostree rebase local:tmpref-rhcos; do sleep 0.1; done |
First, given a build like: https://releases-art-rhcos.svc.ci.openshift.org/art/storage/releases/rhcos-4.3/43.81.201911081536.0/x86_64/meta.json | |
Download the qemu image: | |
``` | |
$ curl -L https://releases-art-rhcos.svc.ci.openshift.org/art/storage/releases/rhcos-4.3/43.81.201911081536.0/x86_64/rhcos-43.81.201911081536.0-qemu.x86_64.qcow2.gz | gunzip > rhcos-43.81.201911081536.0-qemu.x86_64.qcow2 | |
``` | |
Download new RPM or other content you want, in this case let's assume it's a new hyperkube we want to drop in `/usr/bin`: | |
`$ curl -L http://example/openshift-hyperkube-4.3.0-201912131246.git.0.09a9468.el8.x86_64.rpm | rpm2cpio | cpio -div` |
Same experimental setup as https://gist.github.com/cgwalters/64752a9544eec662e056f8387a3cefc2#file-ssd-test-md
Based on
[global]
bs=2000
ioengine=sync
iodepth=1
runtime=60
directory=/mnt
First you'll need an initialized cosa workdir:
cosa init https://github.com/coreos/fedora-coreos-config
Now, we need the previous build data: coreos/coreos-assembler#1013
But we can hand-roll that too:
$ buildid=31.20191219.20.1
``` | |
diff --git a/data/data/gcp/main.tf b/data/data/gcp/main.tf | |
index ec683906e..9268f5244 100644 | |
--- a/data/data/gcp/main.tf | |
+++ b/data/data/gcp/main.tf | |
@@ -98,4 +98,6 @@ resource "google_compute_image" "cluster" { | |
raw_disk { | |
source = var.gcp_image_uri | |
} | |
+ |
From f5771ce23b71492167ce7ce592a369f1a1a5d0c9 Mon Sep 17 00:00:00 2001 | |
From: Colin Walters <[email protected]> | |
Date: Wed, 11 Mar 2020 19:20:51 +0000 | |
Subject: [PATCH] mantle: Really avoid crashing if no port is set for SSH | |
Actually tested this time. | |
--- | |
mantle/platform/cluster.go | 4 +++- | |
1 file changed, 3 insertions(+), 1 deletion(-) |