diff options
Diffstat (limited to 'mirrors/utils.py')
| -rw-r--r-- | mirrors/utils.py | 78 |
1 files changed, 53 insertions, 25 deletions
diff --git a/mirrors/utils.py b/mirrors/utils.py index 32fa3587..3ab176b3 100644 --- a/mirrors/utils.py +++ b/mirrors/utils.py @@ -1,13 +1,14 @@ from datetime import timedelta from django.db.models import Avg, Count, Max, Min, StdDev +from django.utils.timezone import now from django_countries.fields import Country -from main.utils import cache_function, utc_now +from main.utils import cache_function, database_vendor from .models import MirrorLog, MirrorProtocol, MirrorUrl -default_cutoff = timedelta(hours=24) +DEFAULT_CUTOFF = timedelta(hours=24) def annotate_url(url, delays): '''Given a MirrorURL object, add a few more attributes to it regarding @@ -30,36 +31,57 @@ def annotate_url(url, delays): @cache_function(123) -def get_mirror_statuses(cutoff=default_cutoff): - cutoff_time = utc_now() - cutoff - protocols = list(MirrorProtocol.objects.filter(is_download=True)) - # I swear, this actually has decent performance... - urls = MirrorUrl.objects.select_related('mirror', 'protocol').filter( +def get_mirror_statuses(cutoff=DEFAULT_CUTOFF, mirror_ids=None): + cutoff_time = now() - cutoff + + valid_urls = MirrorUrl.objects.filter( mirror__active=True, mirror__public=True, - protocol__in=protocols, - logs__check_time__gte=cutoff_time).annotate( + logs__check_time__gte=cutoff_time).distinct() + + if mirror_ids: + valid_urls = valid_urls.filter(mirror_id__in=mirror_ids) + + url_data = MirrorUrl.objects.values('id', 'mirror_id').filter( + id__in=valid_urls, logs__check_time__gte=cutoff_time).annotate( check_count=Count('logs'), success_count=Count('logs__duration'), last_sync=Max('logs__last_sync'), last_check=Max('logs__check_time'), - duration_avg=Avg('logs__duration'), - duration_stddev=StdDev('logs__duration')) + duration_avg=Avg('logs__duration')) + + vendor = database_vendor(MirrorUrl) + if vendor != 'sqlite': + url_data = url_data.annotate(duration_stddev=StdDev('logs__duration')) + + urls = MirrorUrl.objects.select_related('mirror', 'protocol').filter( + id__in=valid_urls).order_by('mirror__id', 'url') # The Django ORM makes it really hard to get actual average delay in the # above query, so run a seperate query for it and we will process the # results here. - times = MirrorLog.objects.filter(is_success=True, last_sync__isnull=False, + times = MirrorLog.objects.values_list( + 'url_id', 'check_time', 'last_sync').filter( + is_success=True, last_sync__isnull=False, check_time__gte=cutoff_time) + if mirror_ids: + times = times.filter(url__mirror_id__in=mirror_ids) delays = {} - for log in times: - delay = log.check_time - log.last_sync - delays.setdefault(log.url_id, []).append(delay) + for url_id, check_time, last_sync in times: + delay = check_time - last_sync + delays.setdefault(url_id, []).append(delay) if urls: + url_data = dict((item['id'], item) for item in url_data) + for url in urls: + for k, v in url_data.get(url.id, {}).items(): + if k not in ('id', 'mirror_id'): + setattr(url, k, v) last_check = max([u.last_check for u in urls]) num_checks = max([u.check_count for u in urls]) - check_info = MirrorLog.objects.filter( - check_time__gte=cutoff_time).aggregate( + check_info = MirrorLog.objects.filter(check_time__gte=cutoff_time) + if mirror_ids: + check_info = check_info.filter(url__mirror_id__in=mirror_ids) + check_info = check_info.aggregate( mn=Min('check_time'), mx=Max('check_time')) if num_checks > 1: check_frequency = (check_info['mx'] - check_info['mn']) \ @@ -72,6 +94,9 @@ def get_mirror_statuses(cutoff=default_cutoff): check_frequency = None for url in urls: + # fake the standard deviation for local testing setups + if vendor == 'sqlite': + setattr(url, 'duration_stddev', 0.0) annotate_url(url, delays) return { @@ -84,28 +109,31 @@ def get_mirror_statuses(cutoff=default_cutoff): @cache_function(117) -def get_mirror_errors(cutoff=default_cutoff): - cutoff_time = utc_now() - cutoff +def get_mirror_errors(cutoff=DEFAULT_CUTOFF, mirror_ids=None): + cutoff_time = now() - cutoff errors = MirrorLog.objects.filter( is_success=False, check_time__gte=cutoff_time, url__mirror__active=True, url__mirror__public=True).values( 'url__url', 'url__country', 'url__protocol__protocol', - 'url__mirror__country', 'error').annotate( + 'url__mirror__tier', 'error').annotate( error_count=Count('error'), last_occurred=Max('check_time') ).order_by('-last_occurred', '-error_count') + + if mirror_ids: + urls = urls.filter(mirror_id__in=mirror_ids) + errors = list(errors) for err in errors: - ctry_code = err['url__country'] or err['url__mirror__country'] - err['country'] = Country(ctry_code) + err['country'] = Country(err['url__country']) return errors @cache_function(295) -def get_mirror_url_for_download(cutoff=default_cutoff): +def get_mirror_url_for_download(cutoff=DEFAULT_CUTOFF): '''Find a good mirror URL to use for package downloads. If we have mirror status data available, it is used to determine a good choice by looking at the last batch of status rows.''' - cutoff_time = utc_now() - cutoff + cutoff_time = now() - cutoff status_data = MirrorLog.objects.filter( check_time__gte=cutoff_time).aggregate( Max('check_time'), Max('last_sync')) @@ -123,7 +151,7 @@ def get_mirror_url_for_download(cutoff=default_cutoff): mirror_urls = MirrorUrl.objects.filter( mirror__public=True, mirror__active=True, protocol__default=True) # look first for a country-agnostic URL, then fall back to any HTTP URL - filtered_urls = mirror_urls.filter(mirror__country='')[:1] + filtered_urls = mirror_urls.filter(country='')[:1] if not filtered_urls: filtered_urls = mirror_urls[:1] if not filtered_urls: |