Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Choose WAL-G backup by start time when cloning and deleting #1

Closed
Show file tree
Hide file tree
Changes from 6 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion postgres-appliance/Dockerfile
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@ FROM $BASE_IMAGE as dependencies-builder

ARG DEMO

ENV WALG_VERSION=v2.0.1
ENV WALG_VERSION=v3.0.3

COPY build_scripts/dependencies.sh /builddeps/

Expand Down
25 changes: 16 additions & 9 deletions postgres-appliance/bootstrap/clone_with_wale.py
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,7 @@

import argparse
import csv
import json
import logging
import os
import re
Expand Down Expand Up @@ -40,14 +41,16 @@ def read_configuration():
return options(args.scope, args.datadir, recovery_target_time, args.dry_run)


def build_wale_command(command, datadir=None, backup=None):
def build_wale_command(command, datadir=None, backup=None, extra_args=None):
cmd = ['wal-g' if os.getenv('USE_WALG_RESTORE') == 'true' else 'wal-e'] + [command]
if command == 'backup-fetch':
if datadir is None or backup is None:
raise Exception("backup-fetch requires datadir and backup arguments")
cmd.extend([datadir, backup])
elif command != 'backup-list':
raise Exception("invalid {0} command {1}".format(cmd[0], command))
if extra_args:
cmd.extend(extra_args)
return cmd


Expand All @@ -57,8 +60,7 @@ def fix_output(output):
started = None
for line in output.decode('utf-8').splitlines():
if not started:
started = re.match(r'^name\s+last_modified\s+', line)
started = re.match(r'^name\s+last_modified\s+', line) or re.match(r'^name\s+modified\s+', line)
started = re.match(r'^(backup_)?name\s+(last_)?modified\s+', line)
Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Make sure this matches up with the fix upstream so that we don't reintroduce the bug: zalando#1026

if started:
line = line.replace(' modified ', ' last_modified ')
if started:
Expand All @@ -70,20 +72,25 @@ def choose_backup(backup_list, recovery_target_time):

match_timestamp = match = None
for backup in backup_list:
last_modified = parse(backup['last_modified'])
last_modified = parse(backup['start_time' if os.getenv('USE_WALG_RESTORE') == 'true' else 'last_modified'])

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hi Simon, should this be finish_time instead of start_time of the backup to compare with the recovery_target_time? Postgres requires that the recovery target "must be after the ending time of the base backup" (https://www.postgresql.org/docs/13/continuous-archiving.html).

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for bringing this up!

The rational behind picking start_time over finish_time is that one backup might have been started first but finished second. I'm actually not sure how likely parallel backups are, if they are possible at all. Please let me know if you know more on this matter!

Either way, after having read the documentation you linked to I think you are absolutely correct. It does not make sense to pick a base backup that has not yet finished before the "recovery target".

272fbf1

if last_modified < recovery_target_time:
if match is None or last_modified > match_timestamp:
match = backup
match_timestamp = last_modified
if match is not None:
return match['name']
return match.get('name', match['backup_name'])


def list_backups(env):
backup_list_cmd = build_wale_command('backup-list')
output = subprocess.check_output(backup_list_cmd, env=env)
reader = csv.DictReader(fix_output(output), dialect='excel-tab')
return list(reader)
if os.getenv('USE_WALG_RESTORE') == 'true':
backup_list_cmd = build_wale_command('backup-list', extra_args=['--detail', '--json'])
output = subprocess.check_output(backup_list_cmd, env=env)
return json.loads(output or 'null')
else:
backup_list_cmd = build_wale_command('backup-list')
output = subprocess.check_output(backup_list_cmd, env=env)
reader = csv.DictReader(fix_output(output), dialect='excel-tab')
return list(reader)


def get_clone_envdir():
Expand Down
40 changes: 21 additions & 19 deletions postgres-appliance/scripts/postgres_backup.sh
Original file line number Diff line number Diff line change
Expand Up @@ -39,29 +39,31 @@ else
POOL_SIZE=(--pool-size "$POOL_SIZE")
fi

BEFORE=""
LEFT=0

NOW=$(date +%s -u)
readonly NOW
while read -r name last_modified rest; do
last_modified=$(date +%s -ud "$last_modified")
if [ $(((NOW-last_modified)/86400)) -ge $DAYS_TO_RETAIN ]; then
if [ -z "$BEFORE" ] || [ "$last_modified" -gt "$BEFORE_TIME" ]; then
BEFORE_TIME=$last_modified
BEFORE=$name

if [[ "$USE_WALG_BACKUP" == "true" ]]; then
AFTER=$((NOW-(DAYS_TO_RETAIN*86400)))
$WAL_E delete --use-sentinel-time --confirm retain FIND_FULL "$DAYS_TO_RETAIN" --after "$(date --iso-8601=seconds -d @"${AFTER}")"
else
BEFORE=""
LEFT=0

while read -r name last_modified rest; do
last_modified=$(date +%s -ud "$last_modified")
if [ $(((NOW-last_modified)/86400)) -ge $DAYS_TO_RETAIN ]; then
if [ -z "$BEFORE" ] || [ "$last_modified" -gt "$BEFORE_TIME" ]; then
BEFORE_TIME=$last_modified
BEFORE=$name
fi
else
# count how many backups will remain after we remove everything up to certain date
((LEFT=LEFT+1))
fi
else
# count how many backups will remain after we remove everything up to certain date
((LEFT=LEFT+1))
fi
done < <($WAL_E backup-list 2> /dev/null | sed '0,/^name\s*\(last_\)\?modified\s*/d')
done < <($WAL_E backup-list 2> /dev/null | sed '0,/^name\s*\(last_\)\?modified\s*/d')

# we want keep at least N backups even if the number of days exceeded
if [ -n "$BEFORE" ] && [ $LEFT -ge $DAYS_TO_RETAIN ]; then
if [[ "$USE_WALG_BACKUP" == "true" ]]; then
$WAL_E delete before FIND_FULL "$BEFORE" --confirm
else
# we want keep at least N backups even if the number of days exceeded
if [ -n "$BEFORE" ] && [ $LEFT -ge $DAYS_TO_RETAIN ]; then
$WAL_E delete --confirm before "$BEFORE"
fi
fi
Expand Down