Bitnami/containers
2
0
Fork 0
mirror of https://github.com/bitnami/containers.git synced 2026-02-28 15:37:26 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix empty member ID (#21)

Browse Source
This commit is contained in:
Christian Köhn
2021-11-18 15:46:48 +01:00
committed by GitHub
parent 540dba5520
commit ac77d60c4f
4 changed files with 85 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
bitnami/etcd/3/debian-10/rootfs/opt/bitnami/scripts/etcd/postunpack.sh
View File

@@ -13,7 +13,7 @@ set -o pipefail
# Load etcd environment variables
. /opt/bitnami/scripts/etcd-env.sh
for dir in "$ETCD_BIN_DIR" "$ETCD_DATA_DIR"; do
for dir in "$ETCD_BIN_DIR" "$ETCD_DATA_DIR" "${ETCD_BASE_DIR}/certs"; do
ensure_dir_exists "$dir"
done
chmod -R g+rwX "$ETCD_DATA_DIR"
chmod -R g+rwX "$ETCD_DATA_DIR" "${ETCD_BASE_DIR}/certs"

7
bitnami/etcd/3/debian-10/rootfs/opt/bitnami/scripts/etcd/prestop.sh
View File

@@ -5,6 +5,7 @@
set -o errexit
set -o pipefail
set -o nounset
# set -o xtrace # Uncomment this line for debugging purposes
# Load libraries
. /opt/bitnami/scripts/libetcd.sh
@@ -12,6 +13,10 @@ set -o nounset
# Load etcd environment settings
. /opt/bitnami/scripts/etcd-env.sh
endpoints="$(etcdctl_get_endpoints true)"
if is_empty_value "${endpoints}"; then
exit 0
fi
read -r -a extra_flags <<< "$(etcdctl_auth_flags)"
extra_flags+=("--endpoints=$(etcdctl_get_endpoints)" "--debug=true")
extra_flags+=("--endpoints=${endpoints}" "--debug=true")
etcdctl member remove "$(cat "${ETCD_DATA_DIR}/member_id")" "${extra_flags[@]}" > "$(dirname "$ETCD_DATA_DIR")/member_removal.log"

10
bitnami/etcd/3/debian-10/rootfs/opt/bitnami/scripts/etcd/run.sh
View File

@@ -20,13 +20,15 @@ set -o pipefail
EXEC="$(command -v etcd)"
! is_empty_value "$ETCD_ROOT_PASSWORD" && unset ETCD_ROOT_PASSWORD
if ! is_empty_value "$ETCD_INITIAL_CLUSTER" && ! is_new_etcd_cluster; then
ETCD_INITIAL_CLUSTER="$(recalculate_initial_cluster)"
export ETCD_INITIAL_CLUSTER
fi
if [[ -f "$ETCD_NEW_MEMBERS_ENV_FILE" ]]; then
debug "Loading env vars of existing cluster"
. "$ETCD_NEW_MEMBERS_ENV_FILE"
else
# We do not rely on the original value of ETCD_INITIAL_CLUSTER even
# when bootstrapping a new cluster since we cannot assume
# that all nodes will come-up healthy
ETCD_INITIAL_CLUSTER="$(recalculate_initial_cluster)"
export ETCD_INITIAL_CLUSTER
fi
info "** Starting etcd **"

88
bitnami/etcd/3/debian-10/rootfs/opt/bitnami/scripts/libetcd.sh
View File

@@ -112,18 +112,29 @@ etcdctl_get_endpoints() {
local -a endpoints=()
local host domain port
ip_has_hostname() {
ip_has_valid_hostname() {
local ip="${1:?ip is required}"
local parent_domain="${1:?parent_domain is required}"
[[ "$(getent hosts "$ip")" != "" ]] && return 0
# 'getent hosts $ip' can return hostnames in 2 different formats:
# POD_NAME.HEADLESS_SVC_DOMAIN.NAMESPACE.svc.cluster.local (using headless service domain)
# 10-237-136-79.SVC_DOMAIN.NAMESPACE.svc.cluster.local (using POD's IP and service domain)
# We need to discad the latter to avoid issues when TLS verification is enabled.
[[ "$(getent hosts $ip)" = *"$parent_domain"* ]] && return 0
return 1
}
hostname_has_ips() {
local hostname="${1:?hostname is required}"
[[ "$(getent ahosts "$hostname")" != "" ]] && return 0
return 1
}
if is_boolean_yes "$ETCD_ON_K8S"; then
# This piece of code assumes this container is used on a K8s environment
# where etcd members are part of a statefulset that uses a headless service
# to create a unique FQDN per member. Under this circustamences, the
# ETCD_ADVERTISE_CLIENT_URLS are created as follows:
# to create a unique FQDN per member. Under these circumstances, the
# ETCD_ADVERTISE_CLIENT_URLS env. variable is created as follows:
# SCHEME://POD_NAME.HEADLESS_SVC_DOMAIN:CLIENT_PORT
#
# Assuming this, we can extract the HEADLESS_SVC_DOMAIN and obtain
@@ -134,15 +145,18 @@ etcdctl_get_endpoints() {
# When ETCD_CLUSTER_DOMAIN is set, we use that value instead of extracting
# it from ETCD_ADVERTISE_CLIENT_URLS
! is_empty_value "$ETCD_CLUSTER_DOMAIN" && domain="$ETCD_CLUSTER_DOMAIN"
wait_for_dns_lookup "$domain" >/dev/null 2>&1
for ip in $(getent ahosts "$domain" | awk '{print $1}' | uniq); do
if retry_while "ip_has_hostname $ip"; then
h="$(getent hosts "$ip" | awk '{print $2}')"
if ! { [[ $only_others = true ]] && [[ "$h" = "$host" ]]; }; then
endpoints+=("${h}:${port}")
# Depending on the K8s distro & the DNS plugin, it might need
# a few seconds to associate the POD(s) IP(s) to the headless svc domain
if retry_while "hostname_has_ips $domain"; then
for ip in $(getent ahosts "$domain" | awk '{print $1}' | uniq); do
if retry_while "ip_has_valid_hostname $ip $domain"; then
h="$(getent hosts "$ip" | awk '{print $2}')"
if ! { [[ $only_others = true ]] && [[ "$h" = "$host" ]]; }; then
endpoints+=("${h}:${port}")
fi
fi
fi
done
done
fi
echo "${endpoints[*]}" | tr ' ' ','
else
echo ""
@@ -186,8 +200,11 @@ etcd_store_member_id() {
read -r -a extra_flags <<< "$(etcdctl_auth_flags)"
extra_flags+=("--endpoints=$(etcdctl_get_endpoints)")
if retry_while "etcdctl ${extra_flags[*]} member list" >/dev/null 2>&1; then
while [[ ! -f "${ETCD_DATA_DIR}/member_id" ]]; do
etcdctl "${extra_flags[@]}" member list | grep -w "$ETCD_ADVERTISE_CLIENT_URLS" | awk -F "," '{ print $1}' > "${ETCD_DATA_DIR}/member_id"
while [[ ! -s "${ETCD_DATA_DIR}/member_id" ]]; do
# We use 'stdbuf' to ensure memory buffers are flushed to disk
# so we reduce the chances that the "member_id" file is not created.
# ref: https://www.gnu.org/software/coreutils/manual/html_node/stdbuf-invocation.html#stdbuf-invocation
stdbuf -oL etcdctl "${extra_flags[@]}" member list | grep -w "$ETCD_ADVERTISE_CLIENT_URLS" | awk -F "," '{ print $1}' > "${ETCD_DATA_DIR}/member_id" || true
done
debug "Stored member ID: $(cat "${ETCD_DATA_DIR}/member_id")"
fi
@@ -286,7 +303,15 @@ is_healthy_etcd_cluster() {
if is_boolean_yes "$ETCD_DISASTER_RECOVERY"; then
if [[ -f "/snapshots/.disaster_recovery" ]]; then
remove_in_file "/snapshots/.disaster_recovery" "$host:$port"
# Remove current node from the ones that need to recover
remove_in_file "/snapshots/.disaster_recovery" "$host:$port" || true
# Remove nodes that do not exist anymore from the ones that need to recover
read -r -a recovery_array <<< $(tr '\n' ' ' < "/snapshots/.disaster_recovery")
for r in "${recovery_array[@]}"; do
if [[ ! " ${endpoints_array[*]} " =~ " $r " ]]; then
remove_in_file "/snapshots/.disaster_recovery" "$r" || true
fi
done
if [[ $(wc -w < "/snapshots/.disaster_recovery") -eq 0 ]]; then
debug "Last member to recover from the disaster!"
rm "/snapshots/.disaster_recovery"
@@ -312,7 +337,7 @@ is_healthy_etcd_cluster() {
}
########################
# Recalculate initial cluster for disaster recovery
# Recalculate initial cluster
# Globals:
# ETCD_*
# Arguments:
@@ -328,7 +353,7 @@ recalculate_initial_cluster() {
read -r -a endpoints_array <<< "$(tr ',;' ' ' <<< "$(etcdctl_get_endpoints)")"
# This piece of code assumes this container is used on a K8s environment
# where etcd members are part of a statefulset that uses a headless service
# to create a unique FQDN per member. Under this circustamences, the
# to create a unique FQDN per member. Under these circumstances, the
# ETCD_INITIAL_ADVERTISE_PEER_URLS are created as follows:
# SCHEME://POD_NAME.HEADLESS_SVC_DOMAIN:PEER_PORT
#
@@ -364,6 +389,7 @@ recalculate_initial_cluster() {
#########################
etcd_initialize() {
local -a extra_flags initial_members
local domain
info "Initializing etcd"
read -r -a initial_members <<< "$(tr ',;' ' ' <<< "$ETCD_INITIAL_CLUSTER")"
@@ -372,6 +398,34 @@ etcd_initialize() {
if [[ ${#initial_members[@]} -gt 1 ]]; then
if is_new_etcd_cluster; then
info "Bootstrapping a new cluster"
if is_boolean_yes "$ETCD_ON_K8S"; then
debug "Waiting for the headless svc domain to have an IP per initial member in the cluster"
if is_empty_value "$ETCD_CLUSTER_DOMAIN"; then
# This piece of code assumes this container is used on a K8s environment
# where etcd members are part of a statefulset that uses a headless service
# to create a unique FQDN per member. Under these circumstances, the
# ETCD_INITIAL_ADVERTISE_PEER_URLS are created as follows:
# SCHEME://POD_NAME.HEADLESS_SVC_DOMAIN:PEER_PORT
#
# Assuming this, we can extract the HEADLESS_SVC_DOMAIN
host="$(parse_uri "$ETCD_INITIAL_ADVERTISE_PEER_URLS" "host")"
domain="${host#"${ETCD_NAME}."}"
else
# When ETCD_CLUSTER_DOMAIN is set, we use that value instead of extracting
# it from ETCD_INITIAL_ADVERTISE_PEER_URLS
domain="$ETCD_CLUSTER_DOMAIN"
fi
hostname_has_N_ips() {
local -r hostname="${1:?hostname is required}"
local -r n=${2:?number of ips is required}
[[ $(getent ahosts "$hostname" | awk '{print $1}' | uniq | wc -l) -eq $n ]] && return 0
return 1
}
if ! retry_while "hostname_has_N_ips $domain ${#initial_members[@]}"; then
error "Headless service domain does not have an IP per initial member in the cluster"
exit 1
fi
fi
else
info "Adding new member to existing cluster"
ensure_dir_exists "$ETCD_DATA_DIR"