forked from kubernetes/kubernetes
-
Notifications
You must be signed in to change notification settings - Fork 0
/
util.sh
executable file
·1482 lines (1327 loc) · 52.7 KB
/
util.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
#!/bin/bash
# Copyright 2014 The Kubernetes Authors All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# A library of helper functions and constant for the local config.
# Use the config file specified in $KUBE_CONFIG_FILE, or default to
# config-default.sh.
KUBE_ROOT=$(dirname "${BASH_SOURCE}")/../..
source "${KUBE_ROOT}/cluster/gce/${KUBE_CONFIG_FILE-"config-default.sh"}"
source "${KUBE_ROOT}/cluster/common.sh"
if [[ "${OS_DISTRIBUTION}" == "debian" || "${OS_DISTRIBUTION}" == "coreos" || "${OS_DISTRIBUTION}" == "trusty" ]]; then
source "${KUBE_ROOT}/cluster/gce/${OS_DISTRIBUTION}/helper.sh"
else
echo "Cannot operate on cluster using os distro: ${OS_DISTRIBUTION}" >&2
exit 1
fi
NODE_INSTANCE_PREFIX="${INSTANCE_PREFIX}-minion"
ALLOCATE_NODE_CIDRS=true
KUBE_PROMPT_FOR_UPDATE=y
KUBE_SKIP_UPDATE=${KUBE_SKIP_UPDATE-"n"}
# Suffix to append to the staging path used for the server tars. Useful if
# multiple versions of the server are being used in the same project
# simultaneously (e.g. on Jenkins).
KUBE_GCS_STAGING_PATH_SUFFIX=${KUBE_GCS_STAGING_PATH_SUFFIX-""}
# How long (in seconds) to wait for cluster initialization.
KUBE_CLUSTER_INITIALIZATION_TIMEOUT=${KUBE_CLUSTER_INITIALIZATION_TIMEOUT:-300}
function join_csv {
local IFS=','; echo "$*";
}
# Verify prereqs
function verify-prereqs {
local cmd
for cmd in gcloud gsutil; do
if ! which "${cmd}" >/dev/null; then
local resp
if [[ "${KUBE_PROMPT_FOR_UPDATE}" == "y" ]]; then
echo "Can't find ${cmd} in PATH. Do you wish to install the Google Cloud SDK? [Y/n]"
read resp
else
resp="y"
fi
if [[ "${resp}" != "n" && "${resp}" != "N" ]]; then
curl https://sdk.cloud.google.com | bash
fi
if ! which "${cmd}" >/dev/null; then
echo "Can't find ${cmd} in PATH, please fix and retry. The Google Cloud " >&2
echo "SDK can be downloaded from https://cloud.google.com/sdk/." >&2
exit 1
fi
fi
done
if [[ "${KUBE_SKIP_UPDATE}" == "y" ]]; then
return
fi
# update and install components as needed
if [[ "${KUBE_PROMPT_FOR_UPDATE}" != "y" ]]; then
gcloud_prompt="-q"
fi
local sudo_prefix=""
if [ ! -w $(dirname `which gcloud`) ]; then
sudo_prefix="sudo"
fi
${sudo_prefix} gcloud ${gcloud_prompt:-} components install alpha || true
${sudo_prefix} gcloud ${gcloud_prompt:-} components install beta || true
${sudo_prefix} gcloud ${gcloud_prompt:-} components update || true
}
# Create a temp dir that'll be deleted at the end of this bash session.
#
# Vars set:
# KUBE_TEMP
function ensure-temp-dir {
if [[ -z ${KUBE_TEMP-} ]]; then
KUBE_TEMP=$(mktemp -d -t kubernetes.XXXXXX)
trap 'rm -rf "${KUBE_TEMP}"' EXIT
fi
}
# Use the gcloud defaults to find the project. If it is already set in the
# environment then go with that.
#
# Vars set:
# PROJECT
# PROJECT_REPORTED
function detect-project () {
if [[ -z "${PROJECT-}" ]]; then
PROJECT=$(gcloud config list project | tail -n 1 | cut -f 3 -d ' ')
fi
if [[ -z "${PROJECT-}" ]]; then
echo "Could not detect Google Cloud Platform project. Set the default project using " >&2
echo "'gcloud config set project <PROJECT>'" >&2
exit 1
fi
if [[ -z "${PROJECT_REPORTED-}" ]]; then
echo "Project: ${PROJECT}" >&2
echo "Zone: ${ZONE}" >&2
PROJECT_REPORTED=true
fi
}
function sha1sum-file() {
if which shasum >/dev/null 2>&1; then
shasum -a1 "$1" | awk '{ print $1 }'
else
sha1sum "$1" | awk '{ print $1 }'
fi
}
function already-staged() {
local -r file=$1
local -r newsum=$2
[[ -e "${file}.uploaded.sha1" ]] || return 1
local oldsum
oldsum=$(cat "${file}.uploaded.sha1")
[[ "${oldsum}" == "${newsum}" ]]
}
# Copy a release tar, if we don't already think it's staged in GCS
function copy-if-not-staged() {
local -r staging_path=$1
local -r gs_url=$2
local -r tar=$3
local -r hash=$4
if already-staged "${tar}" "${hash}"; then
echo "+++ $(basename ${tar}) already staged ('rm ${tar}.uploaded.sha1' to force)"
else
echo "${hash}" > "${tar}.sha1"
gsutil -m -q -h "Cache-Control:private, max-age=0" cp "${tar}" "${tar}.sha1" "${staging_path}"
gsutil -m acl ch -g all:R "${gs_url}" "${gs_url}.sha1" >/dev/null 2>&1
echo "${hash}" > "${tar}.uploaded.sha1"
echo "+++ $(basename ${tar}) uploaded (sha1 = ${hash})"
fi
}
# Take the local tar files and upload them to Google Storage. They will then be
# downloaded by the master as part of the start up script for the master.
# If running on Ubuntu trusty, we also pack the dir cluster/gce/trusty/kube-manifest
# and upload it to Google Storage.
#
# Assumed vars:
# PROJECT
# SERVER_BINARY_TAR
# SALT_TAR
# KUBE_MANIFESTS_TAR
# Vars set:
# SERVER_BINARY_TAR_URL
# SERVER_BINARY_TAR_HASH
# SALT_TAR_URL
# SALT_TAR_HASH
# KUBE_MANIFESTS_TAR_URL
# KUBE_MANIFESTS_TAR_HASH
function upload-server-tars() {
SERVER_BINARY_TAR_URL=
SERVER_BINARY_TAR_HASH=
SALT_TAR_URL=
SALT_TAR_HASH=
KUBE_MANIFESTS_TAR_URL=
KUBE_MANIFESTS_TAR_HASH=
local project_hash
if which md5 > /dev/null 2>&1; then
project_hash=$(md5 -q -s "$PROJECT")
else
project_hash=$(echo -n "$PROJECT" | md5sum | awk '{ print $1 }')
fi
# This requires 1 million projects before the probability of collision is 50%
# that's probably good enough for now :P
project_hash=${project_hash:0:10}
local -r staging_bucket="gs://kubernetes-staging-${project_hash}"
# Ensure the bucket is created
if ! gsutil ls "$staging_bucket" > /dev/null 2>&1 ; then
echo "Creating $staging_bucket"
gsutil mb "${staging_bucket}"
fi
local -r staging_path="${staging_bucket}/devel${KUBE_GCS_STAGING_PATH_SUFFIX}"
SERVER_BINARY_TAR_HASH=$(sha1sum-file "${SERVER_BINARY_TAR}")
SALT_TAR_HASH=$(sha1sum-file "${SALT_TAR}")
echo "+++ Staging server tars to Google Storage: ${staging_path}"
local server_binary_gs_url="${staging_path}/${SERVER_BINARY_TAR##*/}"
local salt_gs_url="${staging_path}/${SALT_TAR##*/}"
copy-if-not-staged "${staging_path}" "${server_binary_gs_url}" "${SERVER_BINARY_TAR}" "${SERVER_BINARY_TAR_HASH}"
copy-if-not-staged "${staging_path}" "${salt_gs_url}" "${SALT_TAR}" "${SALT_TAR_HASH}"
# Convert from gs:// URL to an https:// URL
SERVER_BINARY_TAR_URL="${server_binary_gs_url/gs:\/\//https://storage.googleapis.com/}"
SALT_TAR_URL="${salt_gs_url/gs:\/\//https://storage.googleapis.com/}"
if [[ "${OS_DISTRIBUTION}" == "trusty" ]]; then
local kube_manifests_gs_url="${staging_path}/${KUBE_MANIFESTS_TAR##*/}"
KUBE_MANIFESTS_TAR_HASH=$(sha1sum-file "${KUBE_MANIFESTS_TAR}")
copy-if-not-staged "${staging_path}" "${kube_manifests_gs_url}" "${KUBE_MANIFESTS_TAR}" "${KUBE_MANIFESTS_TAR_HASH}"
# Convert from gs:// URL to an https:// URL
KUBE_MANIFESTS_TAR_URL="${kube_manifests_gs_url/gs:\/\//https://storage.googleapis.com/}"
fi
}
# Detect minions created in the minion group
#
# Assumed vars:
# NODE_INSTANCE_PREFIX
# Vars set:
# NODE_NAMES
# INSTANCE_GROUPS
function detect-node-names {
detect-project
INSTANCE_GROUPS=()
INSTANCE_GROUPS+=($(gcloud compute instance-groups managed list --zone "${ZONE}" --project "${PROJECT}" | grep ${NODE_INSTANCE_PREFIX} | cut -f1 -d" " || true))
NODE_NAMES=()
if [[ -n "${INSTANCE_GROUPS[@]:-}" ]]; then
for group in "${INSTANCE_GROUPS[@]}"; do
NODE_NAMES+=($(gcloud compute instance-groups managed list-instances \
"${group}" --zone "${ZONE}" --project "${PROJECT}" \
--format=yaml | grep instance: | cut -d ' ' -f 2))
done
echo "INSTANCE_GROUPS=${INSTANCE_GROUPS[*]}" >&2
echo "NODE_NAMES=${NODE_NAMES[*]}" >&2
else
echo "INSTANCE_GROUPS=" >&2
echo "NODE_NAMES=" >&2
fi
}
# Detect the information about the minions
#
# Assumed vars:
# ZONE
# Vars set:
# NODE_NAMES
# KUBE_NODE_IP_ADDRESSES (array)
function detect-nodes () {
detect-project
detect-node-names
KUBE_NODE_IP_ADDRESSES=()
for (( i=0; i<${#NODE_NAMES[@]}; i++)); do
local node_ip=$(gcloud compute instances describe --project "${PROJECT}" --zone "${ZONE}" \
"${NODE_NAMES[$i]}" --fields networkInterfaces[0].accessConfigs[0].natIP \
--format=text | awk '{ print $2 }')
if [[ -z "${node_ip-}" ]] ; then
echo "Did not find ${NODE_NAMES[$i]}" >&2
else
echo "Found ${NODE_NAMES[$i]} at ${minion_ip}"
KUBE_NODE_IP_ADDRESSES+=("${minion_ip}")
fi
done
if [[ -z "${KUBE_NODE_IP_ADDRESSES-}" ]]; then
echo "Could not detect Kubernetes minion nodes. Make sure you've launched a cluster with 'kube-up.sh'" >&2
exit 1
fi
}
# Detect the IP for the master
#
# Assumed vars:
# MASTER_NAME
# ZONE
# Vars set:
# KUBE_MASTER
# KUBE_MASTER_IP
function detect-master () {
detect-project
KUBE_MASTER=${MASTER_NAME}
if [[ -z "${KUBE_MASTER_IP-}" ]]; then
KUBE_MASTER_IP=$(gcloud compute instances describe --project "${PROJECT}" --zone "${ZONE}" \
"${MASTER_NAME}" --fields networkInterfaces[0].accessConfigs[0].natIP \
--format=text | awk '{ print $2 }')
fi
if [[ -z "${KUBE_MASTER_IP-}" ]]; then
echo "Could not detect Kubernetes master node. Make sure you've launched a cluster with 'kube-up.sh'" >&2
exit 1
fi
echo "Using master: $KUBE_MASTER (external IP: $KUBE_MASTER_IP)"
}
# Wait for background jobs to finish. Exit with
# an error status if any of the jobs failed.
function wait-for-jobs {
local fail=0
local job
for job in $(jobs -p); do
wait "${job}" || fail=$((fail + 1))
done
if (( fail != 0 )); then
echo -e "${color_red}${fail} commands failed. Exiting.${color_norm}" >&2
# Ignore failures for now.
# exit 2
fi
}
# Robustly try to create a static ip.
# $1: The name of the ip to create
# $2: The name of the region to create the ip in.
function create-static-ip {
detect-project
local attempt=0
local REGION="$2"
while true; do
if ! gcloud compute addresses create "$1" \
--project "${PROJECT}" \
--region "${REGION}" -q > /dev/null; then
if (( attempt > 4 )); then
echo -e "${color_red}Failed to create static ip $1 ${color_norm}" >&2
exit 2
fi
attempt=$(($attempt+1))
echo -e "${color_yellow}Attempt $attempt failed to create static ip $1. Retrying.${color_norm}" >&2
sleep $(($attempt * 5))
else
break
fi
done
}
# Robustly try to create a firewall rule.
# $1: The name of firewall rule.
# $2: IP ranges.
# $3: Target tags for this firewall rule.
function create-firewall-rule {
detect-project
local attempt=0
while true; do
if ! gcloud compute firewall-rules create "$1" \
--project "${PROJECT}" \
--network "${NETWORK}" \
--source-ranges "$2" \
--target-tags "$3" \
--allow tcp,udp,icmp,esp,ah,sctp; then
if (( attempt > 4 )); then
echo -e "${color_red}Failed to create firewall rule $1 ${color_norm}" >&2
exit 2
fi
echo -e "${color_yellow}Attempt $(($attempt+1)) failed to create firewall rule $1. Retrying.${color_norm}" >&2
attempt=$(($attempt+1))
sleep $(($attempt * 5))
else
break
fi
done
}
# $1: version (required)
function get-template-name-from-version {
# trim template name to pass gce name validation
echo "${NODE_INSTANCE_PREFIX}-template-${1}" | cut -c 1-63 | sed 's/[\.\+]/-/g;s/-*$//g'
}
# Robustly try to create an instance template.
# $1: The name of the instance template.
# $2: The scopes flag.
# $3: The minion start script metadata from file.
# $4: The kube-env metadata.
# $5 and others: Additional user defined metadata.
function create-node-template {
detect-project
local template_name="$1"
# First, ensure the template doesn't exist.
# TODO(zmerlynn): To make this really robust, we need to parse the output and
# add retries. Just relying on a non-zero exit code doesn't
# distinguish an ephemeral failed call from a "not-exists".
if gcloud compute instance-templates describe "$template_name" --project "${PROJECT}" &>/dev/null; then
echo "Instance template ${1} already exists; deleting." >&2
if ! gcloud compute instance-templates delete "$template_name" --project "${PROJECT}" &>/dev/null; then
echo -e "${color_yellow}Failed to delete existing instance template${color_norm}" >&2
exit 2
fi
fi
local attempt=1
local preemptible_minions=""
if [[ "${PREEMPTIBLE_NODE}" == "true" ]]; then
preemptible_minions="--preemptible --maintenance-policy TERMINATE"
fi
while true; do
echo "Attempt ${attempt} to create ${1}" >&2
if ! gcloud compute instance-templates create "$template_name" \
--project "${PROJECT}" \
--machine-type "${NODE_SIZE}" \
--boot-disk-type "${NODE_DISK_TYPE}" \
--boot-disk-size "${NODE_DISK_SIZE}" \
--image-project="${NODE_IMAGE_PROJECT}" \
--image "${NODE_IMAGE}" \
--tags "${NODE_TAG}" \
--network "${NETWORK}" \
${preemptible_minions} \
$2 \
--can-ip-forward \
--metadata-from-file $(echo ${@:3} | tr ' ' ',') >&2; then
if (( attempt > 5 )); then
echo -e "${color_red}Failed to create instance template $template_name ${color_norm}" >&2
exit 2
fi
echo -e "${color_yellow}Attempt ${attempt} failed to create instance template $template_name. Retrying.${color_norm}" >&2
attempt=$(($attempt+1))
sleep $(($attempt * 5))
else
break
fi
done
}
# Robustly try to add metadata on an instance.
# $1: The name of the instance.
# $2...$n: The metadata key=value pairs to add.
function add-instance-metadata {
local -r instance=$1
shift 1
local -r kvs=( "$@" )
detect-project
local attempt=0
while true; do
if ! gcloud compute instances add-metadata "${instance}" \
--project "${PROJECT}" \
--zone "${ZONE}" \
--metadata "${kvs[@]}"; then
if (( attempt > 5 )); then
echo -e "${color_red}Failed to add instance metadata in ${instance} ${color_norm}" >&2
exit 2
fi
echo -e "${color_yellow}Attempt $(($attempt+1)) failed to add metadata in ${instance}. Retrying.${color_norm}" >&2
attempt=$(($attempt+1))
sleep $((5 * $attempt))
else
break
fi
done
}
# Robustly try to add metadata on an instance, from a file.
# $1: The name of the instance.
# $2...$n: The metadata key=file pairs to add.
function add-instance-metadata-from-file {
local -r instance=$1
shift 1
local -r kvs=( "$@" )
detect-project
local attempt=0
while true; do
echo "${kvs[@]}"
if ! gcloud compute instances add-metadata "${instance}" \
--project "${PROJECT}" \
--zone "${ZONE}" \
--metadata-from-file "$(join_csv ${kvs[@]})"; then
if (( attempt > 5 )); then
echo -e "${color_red}Failed to add instance metadata in ${instance} ${color_norm}" >&2
exit 2
fi
echo -e "${color_yellow}Attempt $(($attempt+1)) failed to add metadata in ${instance}. Retrying.${color_norm}" >&2
attempt=$(($attempt+1))
sleep $(($attempt * 5))
else
break
fi
done
}
# Quote something appropriate for a yaml string.
#
# TODO(zmerlynn): Note that this function doesn't so much "quote" as
# "strip out quotes", and we really should be using a YAML library for
# this, but PyYAML isn't shipped by default, and *rant rant rant ... SIGH*
function yaml-quote {
echo "'$(echo "${@}" | sed -e "s/'/''/g")'"
}
function write-master-env {
# If the user requested that the master be part of the cluster, set the
# environment variable to program the master kubelet to register itself.
if [[ "${REGISTER_MASTER_KUBELET:-}" == "true" ]]; then
KUBELET_APISERVER="${MASTER_NAME}"
fi
build-kube-env true "${KUBE_TEMP}/master-kube-env.yaml"
}
function write-node-env {
build-kube-env false "${KUBE_TEMP}/node-kube-env.yaml"
}
# Create certificate pairs for the cluster.
# $1: The public IP for the master.
#
# These are used for static cert distribution (e.g. static clustering) at
# cluster creation time. This will be obsoleted once we implement dynamic
# clustering.
#
# The following certificate pairs are created:
#
# - ca (the cluster's certificate authority)
# - server
# - kubelet
# - kubecfg (for kubectl)
#
# TODO(roberthbailey): Replace easyrsa with a simple Go program to generate
# the certs that we need.
#
# Assumed vars
# KUBE_TEMP
#
# Vars set:
# CERT_DIR
# CA_CERT_BASE64
# MASTER_CERT_BASE64
# MASTER_KEY_BASE64
# KUBELET_CERT_BASE64
# KUBELET_KEY_BASE64
# KUBECFG_CERT_BASE64
# KUBECFG_KEY_BASE64
function create-certs {
local -r cert_ip="${1}"
local octets=($(echo "$SERVICE_CLUSTER_IP_RANGE" | sed -e 's|/.*||' -e 's/\./ /g'))
((octets[3]+=1))
local -r service_ip=$(echo "${octets[*]}" | sed 's/ /./g')
local -r sans="IP:${cert_ip},IP:${service_ip},DNS:kubernetes,DNS:kubernetes.default,DNS:kubernetes.default.svc,DNS:kubernetes.default.svc.${DNS_DOMAIN},DNS:${MASTER_NAME}"
local -r cert_create_debug_output=$(mktemp "${KUBE_TEMP}/cert_create_debug_output.XXX")
# Note: This was heavily cribbed from make-ca-cert.sh
(set -x
cd "${KUBE_TEMP}"
curl -L -O --connect-timeout 20 --retry 6 --retry-delay 2 https://storage.googleapis.com/kubernetes-release/easy-rsa/easy-rsa.tar.gz
tar xzf easy-rsa.tar.gz
cd easy-rsa-master/easyrsa3
./easyrsa init-pki
./easyrsa --batch "--req-cn=${cert_ip}@$(date +%s)" build-ca nopass
./easyrsa --subject-alt-name="${sans}" build-server-full "${MASTER_NAME}" nopass
./easyrsa build-client-full kubelet nopass
./easyrsa build-client-full kubecfg nopass) &>${cert_create_debug_output} || {
# If there was an error in the subshell, just die.
# TODO(roberthbailey): add better error handling here
cat "${cert_create_debug_output}" >&2
echo "=== Failed to generate certificates: Aborting ===" >&2
exit 2
}
CERT_DIR="${KUBE_TEMP}/easy-rsa-master/easyrsa3"
# By default, linux wraps base64 output every 76 cols, so we use 'tr -d' to remove whitespaces.
# Note 'base64 -w0' doesn't work on Mac OS X, which has different flags.
CA_CERT_BASE64=$(cat "${CERT_DIR}/pki/ca.crt" | base64 | tr -d '\r\n')
MASTER_CERT_BASE64=$(cat "${CERT_DIR}/pki/issued/${MASTER_NAME}.crt" | base64 | tr -d '\r\n')
MASTER_KEY_BASE64=$(cat "${CERT_DIR}/pki/private/${MASTER_NAME}.key" | base64 | tr -d '\r\n')
KUBELET_CERT_BASE64=$(cat "${CERT_DIR}/pki/issued/kubelet.crt" | base64 | tr -d '\r\n')
KUBELET_KEY_BASE64=$(cat "${CERT_DIR}/pki/private/kubelet.key" | base64 | tr -d '\r\n')
KUBECFG_CERT_BASE64=$(cat "${CERT_DIR}/pki/issued/kubecfg.crt" | base64 | tr -d '\r\n')
KUBECFG_KEY_BASE64=$(cat "${CERT_DIR}/pki/private/kubecfg.key" | base64 | tr -d '\r\n')
}
# Instantiate a kubernetes cluster
#
# Assumed vars
# KUBE_ROOT
# <Various vars set in config file>
function kube-up {
ensure-temp-dir
detect-project
load-or-gen-kube-basicauth
load-or-gen-kube-bearertoken
# Make sure we have the tar files staged on Google Storage
find-release-tars
upload-server-tars
local running_in_terminal=false
# May be false if tty is not allocated (for example with ssh -T).
if [ -t 1 ]; then
running_in_terminal=true
fi
if [[ ${running_in_terminal} == "true" || ${KUBE_UP_AUTOMATIC_CLEANUP} == "true" ]]; then
if ! check-resources; then
local run_kube_down="n"
echo "${KUBE_RESOURCE_FOUND} found." >&2
# Get user input only if running in terminal.
if [[ ${running_in_terminal} == "true" && ${KUBE_UP_AUTOMATIC_CLEANUP} == "false" ]]; then
read -p "Would you like to shut down the old cluster (call kube-down)? [y/N] " run_kube_down
fi
if [[ ${run_kube_down} == "y" || ${run_kube_down} == "Y" || ${KUBE_UP_AUTOMATIC_CLEANUP} == "true" ]]; then
echo "... calling kube-down" >&2
kube-down
fi
fi
fi
if ! gcloud compute networks --project "${PROJECT}" describe "${NETWORK}" &>/dev/null; then
echo "Creating new network: ${NETWORK}"
# The network needs to be created synchronously or we have a race. The
# firewalls can be added concurrent with instance creation.
gcloud compute networks create --project "${PROJECT}" "${NETWORK}" --range "10.240.0.0/16"
fi
if ! gcloud compute firewall-rules --project "${PROJECT}" describe "${NETWORK}-default-internal" &>/dev/null; then
gcloud compute firewall-rules create "${NETWORK}-default-internal" \
--project "${PROJECT}" \
--network "${NETWORK}" \
--source-ranges "10.0.0.0/8" \
--allow "tcp:1-65535,udp:1-65535,icmp" &
fi
if ! gcloud compute firewall-rules describe --project "${PROJECT}" "${NETWORK}-default-ssh" &>/dev/null; then
gcloud compute firewall-rules create "${NETWORK}-default-ssh" \
--project "${PROJECT}" \
--network "${NETWORK}" \
--source-ranges "0.0.0.0/0" \
--allow "tcp:22" &
fi
echo "Starting master and configuring firewalls"
gcloud compute firewall-rules create "${MASTER_NAME}-https" \
--project "${PROJECT}" \
--network "${NETWORK}" \
--target-tags "${MASTER_TAG}" \
--allow tcp:443 &
# We have to make sure the disk is created before creating the master VM, so
# run this in the foreground.
gcloud compute disks create "${MASTER_NAME}-pd" \
--project "${PROJECT}" \
--zone "${ZONE}" \
--type "${MASTER_DISK_TYPE}" \
--size "${MASTER_DISK_SIZE}"
# Create disk for cluster registry if enabled
if [[ "${ENABLE_CLUSTER_REGISTRY}" == true && -n "${CLUSTER_REGISTRY_DISK}" ]]; then
gcloud compute disks create "${CLUSTER_REGISTRY_DISK}" \
--project "${PROJECT}" \
--zone "${ZONE}" \
--type "${CLUSTER_REGISTRY_DISK_TYPE_GCE}" \
--size "${CLUSTER_REGISTRY_DISK_SIZE}" &
fi
# Generate a bearer token for this cluster. We push this separately
# from the other cluster variables so that the client (this
# computer) can forget it later. This should disappear with
# http://issue.k8s.io/3168
KUBELET_TOKEN=$(dd if=/dev/urandom bs=128 count=1 2>/dev/null | base64 | tr -d "=+/" | dd bs=32 count=1 2>/dev/null)
KUBE_PROXY_TOKEN=$(dd if=/dev/urandom bs=128 count=1 2>/dev/null | base64 | tr -d "=+/" | dd bs=32 count=1 2>/dev/null)
# Reserve the master's IP so that it can later be transferred to another VM
# without disrupting the kubelets. IPs are associated with regions, not zones,
# so extract the region name, which is the same as the zone but with the final
# dash and characters trailing the dash removed.
local REGION=${ZONE%-*}
create-static-ip "${MASTER_NAME}-ip" "${REGION}"
MASTER_RESERVED_IP=$(gcloud compute addresses describe "${MASTER_NAME}-ip" \
--project "${PROJECT}" \
--region "${REGION}" -q --format yaml | awk '/^address:/ { print $2 }')
create-certs "${MASTER_RESERVED_IP}"
create-master-instance "${MASTER_RESERVED_IP}" &
# Create a single firewall rule for all minions.
create-firewall-rule "${NODE_TAG}-all" "${CLUSTER_IP_RANGE}" "${NODE_TAG}" &
# Report logging choice (if any).
if [[ "${ENABLE_NODE_LOGGING-}" == "true" ]]; then
echo "+++ Logging using Fluentd to ${LOGGING_DESTINATION:-unknown}"
fi
# Wait for last batch of jobs
wait-for-jobs
echo "Creating minions."
# TODO(zmerlynn): Refactor setting scope flags.
local scope_flags=
if [ -n "${NODE_SCOPES}" ]; then
scope_flags="--scopes ${NODE_SCOPES}"
else
scope_flags="--no-scopes"
fi
write-node-env
local template_name="${NODE_INSTANCE_PREFIX}-template"
create-node-instance-template $template_name
local defaulted_max_instances_per_mig=${MAX_INSTANCES_PER_MIG:-500}
if [[ ${defaulted_max_instances_per_mig} -le "0" ]]; then
echo "MAX_INSTANCES_PER_MIG cannot be negative. Assuming default 500"
defaulted_max_instances_per_mig=500
fi
local num_migs=$(((${NUM_NODES} + ${defaulted_max_instances_per_mig} - 1) / ${defaulted_max_instances_per_mig}))
local instances_per_mig=$(((${NUM_NODES} + ${num_migs} - 1) / ${num_migs}))
local last_mig_size=$((${NUM_NODES} - (${num_migs} - 1) * ${instances_per_mig}))
#TODO: parallelize this loop to speed up the process
for i in $(seq $((${num_migs} - 1))); do
gcloud compute instance-groups managed \
create "${NODE_INSTANCE_PREFIX}-group-$i" \
--project "${PROJECT}" \
--zone "${ZONE}" \
--base-instance-name "${NODE_INSTANCE_PREFIX}" \
--size "${instances_per_mig}" \
--template "$template_name" || true;
gcloud compute instance-groups managed wait-until-stable \
"${NODE_INSTANCE_PREFIX}-group-$i" \
--zone "${ZONE}" \
--project "${PROJECT}" || true;
done
# TODO: We don't add a suffix for the last group to keep backward compatibility when there's only one MIG.
# We should change it at some point, but note #18545 when changing this.
gcloud compute instance-groups managed \
create "${NODE_INSTANCE_PREFIX}-group" \
--project "${PROJECT}" \
--zone "${ZONE}" \
--base-instance-name "${NODE_INSTANCE_PREFIX}" \
--size "${last_mig_size}" \
--template "$template_name" || true;
gcloud compute instance-groups managed wait-until-stable \
"${NODE_INSTANCE_PREFIX}-group" \
--zone "${ZONE}" \
--project "${PROJECT}" || true;
detect-node-names
detect-master
# Create autoscaler for nodes if requested
if [[ "${ENABLE_NODE_AUTOSCALER}" == "true" ]]; then
METRICS=""
# Current usage
METRICS+="--custom-metric-utilization metric=custom.cloudmonitoring.googleapis.com/kubernetes.io/cpu/node_utilization,"
METRICS+="utilization-target=${TARGET_NODE_UTILIZATION},utilization-target-type=GAUGE "
METRICS+="--custom-metric-utilization metric=custom.cloudmonitoring.googleapis.com/kubernetes.io/memory/node_utilization,"
METRICS+="utilization-target=${TARGET_NODE_UTILIZATION},utilization-target-type=GAUGE "
# Reservation
METRICS+="--custom-metric-utilization metric=custom.cloudmonitoring.googleapis.com/kubernetes.io/cpu/node_reservation,"
METRICS+="utilization-target=${TARGET_NODE_UTILIZATION},utilization-target-type=GAUGE "
METRICS+="--custom-metric-utilization metric=custom.cloudmonitoring.googleapis.com/kubernetes.io/memory/node_reservation,"
METRICS+="utilization-target=${TARGET_NODE_UTILIZATION},utilization-target-type=GAUGE "
echo "Creating node autoscalers."
local max_instances_per_mig=$(((${AUTOSCALER_MAX_NODES} + ${num_migs} - 1) / ${num_migs}))
local last_max_instances=$((${AUTOSCALER_MAX_NODES} - (${num_migs} - 1) * ${max_instances_per_mig}))
local min_instances_per_mig=$(((${AUTOSCALER_MIN_NODES} + ${num_migs} - 1) / ${num_migs}))
local last_min_instances=$((${AUTOSCALER_MIN_NODES} - (${num_migs} - 1) * ${min_instances_per_mig}))
for i in $(seq $((${num_migs} - 1))); do
gcloud compute instance-groups managed set-autoscaling "${NODE_INSTANCE_PREFIX}-group-$i" --zone "${ZONE}" --project "${PROJECT}" \
--min-num-replicas "${min_instances_per_mig}" --max-num-replicas "${max_instances_per_mig}" ${METRICS} || true
done
gcloud compute instance-groups managed set-autoscaling "${NODE_INSTANCE_PREFIX}-group" --zone "${ZONE}" --project "${PROJECT}" \
--min-num-replicas "${last_min_instances}" --max-num-replicas "${last_max_instances}" ${METRICS} || true
fi
echo "Waiting up to ${KUBE_CLUSTER_INITIALIZATION_TIMEOUT} seconds for cluster initialization."
echo
echo " This will continually check to see if the API for kubernetes is reachable."
echo " This may time out if there was some uncaught error during start up."
echo
# curl in mavericks is borked.
secure=""
if which sw_vers >& /dev/null; then
if [[ $(sw_vers | grep ProductVersion | awk '{print $2}') = "10.9."* ]]; then
secure="--insecure"
fi
fi
local start_time=$(date +%s)
until curl --cacert "${CERT_DIR}/pki/ca.crt" \
-H "Authorization: Bearer ${KUBE_BEARER_TOKEN}" \
${secure} \
--max-time 5 --fail --output /dev/null --silent \
"https://${KUBE_MASTER_IP}/api/v1/pods"; do
local elapsed=$(($(date +%s) - ${start_time}))
if [[ ${elapsed} -gt ${KUBE_CLUSTER_INITIALIZATION_TIMEOUT} ]]; then
echo -e "${color_red}Cluster failed to initialize within ${KUBE_CLUSTER_INITIALIZATION_TIMEOUT} seconds.${color_norm}" >&2
exit 2
fi
printf "."
sleep 2
done
echo "Kubernetes cluster created."
export KUBE_CERT="${CERT_DIR}/pki/issued/kubecfg.crt"
export KUBE_KEY="${CERT_DIR}/pki/private/kubecfg.key"
export CA_CERT="${CERT_DIR}/pki/ca.crt"
export CONTEXT="${PROJECT}_${INSTANCE_PREFIX}"
(
umask 077
create-kubeconfig
)
# ensures KUBECONFIG is set
get-kubeconfig-basicauth
echo
echo -e "${color_green}Kubernetes cluster is running. The master is running at:"
echo
echo -e "${color_yellow} https://${KUBE_MASTER_IP}"
echo
echo -e "${color_green}The user name and password to use is located in ${KUBECONFIG}.${color_norm}"
echo
}
# Delete a kubernetes cluster. This is called from test-teardown.
#
# Assumed vars:
# MASTER_NAME
# NODE_INSTANCE_PREFIX
# ZONE
# This function tears down cluster resources 10 at a time to avoid issuing too many
# API calls and exceeding API quota. It is important to bring down the instances before bringing
# down the firewall rules and routes.
function kube-down {
detect-project
detect-node-names # For INSTANCE_GROUPS
echo "Bringing down cluster"
set +e # Do not stop on error
# Delete autoscaler for nodes if present. We assume that all or none instance groups have an autoscaler
local autoscaler
autoscaler=( $(gcloud compute instance-groups managed list --zone "${ZONE}" --project "${PROJECT}" \
| grep "${NODE_INSTANCE_PREFIX}-group" \
| awk '{print $7}') )
if [[ "${autoscaler:-}" == "yes" ]]; then
for group in ${INSTANCE_GROUPS[@]:-}; do
gcloud compute instance-groups managed stop-autoscaling "${group}" --zone "${ZONE}" --project "${PROJECT}"
done
fi
# Get the name of the managed instance group template before we delete the
# managed instange group. (The name of the managed instnace group template may
# change during a cluster upgrade.)
local template=$(get-template "${PROJECT}" "${ZONE}" "${NODE_INSTANCE_PREFIX}-group")
# The gcloud APIs don't return machine parseable error codes/retry information. Therefore the best we can
# do is parse the output and special case particular responses we are interested in.
for group in ${INSTANCE_GROUPS[@]:-}; do
if gcloud compute instance-groups managed describe "${group}" --project "${PROJECT}" --zone "${ZONE}" &>/dev/null; then
deleteCmdOutput=$(gcloud compute instance-groups managed delete --zone "${ZONE}" \
--project "${PROJECT}" \
--quiet \
"${group}")
if [[ "$deleteCmdOutput" != "" ]]; then
# Managed instance group deletion is done asynchronously, we must wait for it to complete, or subsequent steps fail
deleteCmdOperationId=$(echo $deleteCmdOutput | grep "Operation:" | sed "s/.*Operation:[[:space:]]*\([^[:space:]]*\).*/\1/g")
if [[ "$deleteCmdOperationId" != "" ]]; then
deleteCmdStatus="PENDING"
while [[ "$deleteCmdStatus" != "DONE" ]]
do
sleep 5
deleteCmdOperationOutput=$(gcloud compute instance-groups managed --zone "${ZONE}" --project "${PROJECT}" get-operation $deleteCmdOperationId)
deleteCmdStatus=$(echo $deleteCmdOperationOutput | grep -i "status:" | sed "s/.*status:[[:space:]]*\([^[:space:]]*\).*/\1/g")
echo "Waiting for MIG deletion to complete. Current status: " $deleteCmdStatus
done
fi
fi
fi
done
if gcloud compute instance-templates describe --project "${PROJECT}" "${template}" &>/dev/null; then
gcloud compute instance-templates delete \
--project "${PROJECT}" \
--quiet \
"${template}"
fi
# First delete the master (if it exists).
if gcloud compute instances describe "${MASTER_NAME}" --zone "${ZONE}" --project "${PROJECT}" &>/dev/null; then
gcloud compute instances delete \
--project "${PROJECT}" \
--quiet \
--delete-disks all \
--zone "${ZONE}" \
"${MASTER_NAME}"
fi
# Delete the master pd (possibly leaked by kube-up if master create failed).
if gcloud compute disks describe "${MASTER_NAME}"-pd --zone "${ZONE}" --project "${PROJECT}" &>/dev/null; then
gcloud compute disks delete \
--project "${PROJECT}" \
--quiet \
--zone "${ZONE}" \
"${MASTER_NAME}"-pd
fi
# Delete disk for cluster registry if enabled
if [[ "${ENABLE_CLUSTER_REGISTRY}" == true && -n "${CLUSTER_REGISTRY_DISK}" ]]; then
if gcloud compute disks describe "${CLUSTER_REGISTRY_DISK}" --zone "${ZONE}" --project "${PROJECT}" &>/dev/null; then
gcloud compute disks delete \
--project "${PROJECT}" \
--quiet \
--zone "${ZONE}" \
"${CLUSTER_REGISTRY_DISK}"
fi
fi
# Find out what minions are running.
local -a minions
minions=( $(gcloud compute instances list \
--project "${PROJECT}" --zone "${ZONE}" \
--regexp "${NODE_INSTANCE_PREFIX}-.+" \
| awk 'NR >= 2 { print $1 }') )
# If any minions are running, delete them in batches.
while (( "${#minions[@]}" > 0 )); do
echo Deleting nodes "${minions[*]::10}"
gcloud compute instances delete \
--project "${PROJECT}" \
--quiet \
--delete-disks boot \
--zone "${ZONE}" \
"${minions[@]::10}"
minions=( "${minions[@]:10}" )
done
# Delete firewall rule for the master.
if gcloud compute firewall-rules describe --project "${PROJECT}" "${MASTER_NAME}-https" &>/dev/null; then
gcloud compute firewall-rules delete \
--project "${PROJECT}" \
--quiet \
"${MASTER_NAME}-https"
fi
# Delete firewall rule for minions.
if gcloud compute firewall-rules describe --project "${PROJECT}" "${NODE_TAG}-all" &>/dev/null; then
gcloud compute firewall-rules delete \
--project "${PROJECT}" \
--quiet \
"${NODE_TAG}-all"
fi
# Delete routes.
local -a routes
# Clean up all routes w/ names like "<cluster-name>-<node-GUID>"
# e.g. "kubernetes-12345678-90ab-cdef-1234-567890abcdef". The name is
# determined by the node controller on the master.
# Note that this is currently a noop, as synchronously deleting the node MIG
# first allows the master to cleanup routes itself.
local TRUNCATED_PREFIX="${INSTANCE_PREFIX:0:26}"
routes=( $(gcloud compute routes list --project "${PROJECT}" \
--regexp "${TRUNCATED_PREFIX}-.{8}-.{4}-.{4}-.{4}-.{12}" | awk 'NR >= 2 { print $1 }') )
while (( "${#routes[@]}" > 0 )); do
echo Deleting routes "${routes[*]::10}"
gcloud compute routes delete \
--project "${PROJECT}" \
--quiet \
"${routes[@]::10}"
routes=( "${routes[@]:10}" )
done
# Delete the master's reserved IP
local REGION=${ZONE%-*}
if gcloud compute addresses describe "${MASTER_NAME}-ip" --region "${REGION}" --project "${PROJECT}" &>/dev/null; then
gcloud compute addresses delete \
--project "${PROJECT}" \
--region "${REGION}" \
--quiet \
"${MASTER_NAME}-ip"
fi
export CONTEXT="${PROJECT}_${INSTANCE_PREFIX}"
clear-kubeconfig
set -e
}
# Gets the instance template for the managed instance group with the provided
# project, zone, and group name. It echos the template name so that the function
# output can be used.
#
# $1: project
# $2: zone
# $3: managed instance group name