rcleere

#!/bin/bash -e

DISKLABEL='local-ssd'
LOCALSSD='/dev/disk/by-id/google-local-ssd'
NVME='/dev/nvme0n1'
MOUNT=/mnt/disks/$DISKLABEL
MKFS='/sbin/mkfs.ext4 -F -E lazy_itable_init=0,lazy_journal_init=0'

dev=""
if [ -e "$NVME" ]; then

rcleere

#!/bin/bash
# Mostly copied from ansible coreos bootstrap
# https://github.com/defunctzombie/ansible-coreos-bootstrap/blob/master/files/bootstrap.sh

set -e

HOME=/home/core

cd

rcleere

import re
import time
import os
from  multiprocessing import Process
import subprocess

IGNORES = ["^loop", "^dm", "^nvme"]

MKFS = ['/sbin/mkfs.ext4', '-F', '-E', 'lazy_itable_init=0,lazy_journal_init=0']

rcleere

We have a user deployed production Kubernetes cluster in GCE that uses Managed Instance Groups for it's worker pools. We have CA installed with the `price` expander. We have 4 zonal MIGS of standard `n1-standard-16` type. We also define 2 other worker pools of `n1-standard-16` type by also specify GPU=`NVIDIA Tesla P100`. In our deployment we had a misconfiguration on the GPU MIGs and we set GPU_COUNT=0. So our worker MIGs look similar to the following, in alphabetical order:

mig-wrk-1234 - US-Central1-A - n1-standard-16, 64GB SSD persistent disk
mig-wrk-abcd - US-Central1-A - n1-standard-16, 64GB SSD persistent disk
mig-wrk-gpu-1234 - US-Central1-C - n1-standard-16, 64GB SSD persistent disk, 0xNVIDIA Tesla P100
mig-wrk-gpu-abcd - US-Central1-F - n1-standard-16, 64GB SSD persistent disk, 0xNVIDIA Tesla P100
mig-wrk-xxxx - US-Central1-A - n1-standard-16, 64GB SSD persistent disk
mig-wrk-yyyy - US-Central1-A - n1-standard-16, 64GB SSD persistent disk

What we were seeing is we created new pods and the CA wante

rcleere

diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c
index 6fb216e..f1a5722 100644
--- a/src/conf/domain_conf.c
+++ b/src/conf/domain_conf.c
@@ -12966,6 +12966,23 @@ virDomainDefParseXML(xmlDocPtr xml,
         VIR_FREE(tmp);
     }

+    /* Extract rlimit_nofiles */
+    n = virXPathULong("string(./rlimit_nofile[1])", ctxt, &count);