Merge "Add json output option to swift-dispersion-report"

This commit is contained in:
Jenkins 2012-02-29 22:09:10 +00:00 committed by Gerrit Code Review
commit 64cdc19490
4 changed files with 117 additions and 62 deletions

View File

@ -14,29 +14,31 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import csv
import os
import socket
from ConfigParser import ConfigParser
from httplib import HTTPException
from optparse import OptionParser
from sys import argv, exit, stdout, stderr
from sys import exit, stdout, stderr
from time import time
from uuid import uuid4
try:
import simplejson as json
except ImportError:
import json
from eventlet import GreenPool, hubs, patcher, sleep, Timeout
from eventlet import GreenPool, hubs, patcher, Timeout
from eventlet.pools import Pool
from swift.common import direct_client
from swift.common.client import ClientException, Connection, get_auth
from swift.common.ring import Ring
from swift.common.utils import compute_eta, get_time_units
from swift.common.utils import compute_eta, get_time_units, TRUE_VALUES
unmounted = []
json_output = False
def get_error_log(prefix):
def error_log(msg_or_exc):
global unmounted
if hasattr(msg_or_exc, 'http_status') and \
@ -98,10 +100,11 @@ def container_dispersion_report(coropool, connpool, account, container_ring,
next_report[0] = time() + 5
eta, eta_unit = compute_eta(begun, containers_queried[0],
containers_listed)
print '\r\x1B[KQuerying containers: %d of %d, %d%s left, %d ' \
'retries' % (containers_queried[0], containers_listed,
round(eta), eta_unit, retries_done[0]),
stdout.flush()
if not json_output:
print '\r\x1B[KQuerying containers: %d of %d, %d%s left, %d ' \
'retries' % (containers_queried[0], containers_listed,
round(eta), eta_unit, retries_done[0]),
stdout.flush()
container_parts = {}
for container in containers:
part, nodes = container_ring.get_nodes(account, container)
@ -114,26 +117,37 @@ def container_dispersion_report(coropool, connpool, account, container_ring,
copies_found = sum(a * b for a, b in enumerate(container_copies_found))
value = 100.0 * copies_found / copies_expected
elapsed, elapsed_unit = get_time_units(time() - begun)
print '\r\x1B[KQueried %d containers for dispersion reporting, ' \
'%d%s, %d retries' % (containers_listed, round(elapsed),
elapsed_unit, retries_done[0])
if containers_listed - distinct_partitions:
print 'There were %d overlapping partitions' % (
containers_listed - distinct_partitions)
if container_copies_found[2]:
print 'There were %d partitions missing one copy.' % \
container_copies_found[2]
if container_copies_found[1]:
print '! There were %d partitions missing two copies.' % \
container_copies_found[1]
if container_copies_found[0]:
print '!!! There were %d partitions missing all copies.' % \
container_copies_found[0]
print '%.02f%% of container copies found (%d of %d)' % (
value, copies_found, copies_expected)
print 'Sample represents %.02f%% of the container partition space' % (
100.0 * distinct_partitions / container_ring.partition_count)
stdout.flush()
if not json_output:
print '\r\x1B[KQueried %d containers for dispersion reporting, ' \
'%d%s, %d retries' % (containers_listed, round(elapsed),
elapsed_unit, retries_done[0])
if containers_listed - distinct_partitions:
print 'There were %d overlapping partitions' % (
containers_listed - distinct_partitions)
if container_copies_found[2]:
print 'There were %d partitions missing one copy.' % \
container_copies_found[2]
if container_copies_found[1]:
print '! There were %d partitions missing two copies.' % \
container_copies_found[1]
if container_copies_found[0]:
print '!!! There were %d partitions missing all copies.' % \
container_copies_found[0]
print '%.02f%% of container copies found (%d of %d)' % (
value, copies_found, copies_expected)
print 'Sample represents %.02f%% of the container partition space' % (
100.0 * distinct_partitions / container_ring.partition_count)
stdout.flush()
return None
else:
return {'retries:': retries_done[0],
'overlapping': containers_listed - distinct_partitions,
'missing_one': container_copies_found[2],
'missing_two': container_copies_found[1],
'missing_all': container_copies_found[0],
'pct_found': value,
'copies_found': copies_found,
'copies_expected': copies_expected}
def object_dispersion_report(coropool, connpool, account, object_ring,
@ -186,9 +200,10 @@ def object_dispersion_report(coropool, connpool, account, object_ring,
next_report[0] = time() + 5
eta, eta_unit = compute_eta(begun, objects_queried[0],
objects_listed)
print '\r\x1B[KQuerying objects: %d of %d, %d%s left, %d ' \
'retries' % (objects_queried[0], objects_listed, round(eta),
eta_unit, retries_done[0]),
if not json_output:
print '\r\x1B[KQuerying objects: %d of %d, %d%s left, %d ' \
'retries' % (objects_queried[0], objects_listed,
round(eta), eta_unit, retries_done[0]),
stdout.flush()
object_parts = {}
for obj in objects:
@ -202,37 +217,56 @@ def object_dispersion_report(coropool, connpool, account, object_ring,
copies_found = sum(a * b for a, b in enumerate(object_copies_found))
value = 100.0 * copies_found / copies_expected
elapsed, elapsed_unit = get_time_units(time() - begun)
print '\r\x1B[KQueried %d objects for dispersion reporting, ' \
'%d%s, %d retries' % (objects_listed, round(elapsed),
elapsed_unit, retries_done[0])
if objects_listed - distinct_partitions:
print 'There were %d overlapping partitions' % (
objects_listed - distinct_partitions)
if object_copies_found[2]:
print 'There were %d partitions missing one copy.' % \
object_copies_found[2]
if object_copies_found[1]:
print '! There were %d partitions missing two copies.' % \
object_copies_found[1]
if object_copies_found[0]:
print '!!! There were %d partitions missing all copies.' % \
object_copies_found[0]
print '%.02f%% of object copies found (%d of %d)' % \
(value, copies_found, copies_expected)
print 'Sample represents %.02f%% of the object partition space' % (
100.0 * distinct_partitions / object_ring.partition_count)
stdout.flush()
if not json_output:
print '\r\x1B[KQueried %d objects for dispersion reporting, ' \
'%d%s, %d retries' % (objects_listed, round(elapsed),
elapsed_unit, retries_done[0])
if objects_listed - distinct_partitions:
print 'There were %d overlapping partitions' % (
objects_listed - distinct_partitions)
if object_copies_found[2]:
print 'There were %d partitions missing one copy.' % \
object_copies_found[2]
if object_copies_found[1]:
print '! There were %d partitions missing two copies.' % \
object_copies_found[1]
if object_copies_found[0]:
print '!!! There were %d partitions missing all copies.' % \
object_copies_found[0]
print '%.02f%% of object copies found (%d of %d)' % \
(value, copies_found, copies_expected)
print 'Sample represents %.02f%% of the object partition space' % (
100.0 * distinct_partitions / object_ring.partition_count)
stdout.flush()
return None
else:
return {'retries:': retries_done[0],
'overlapping': objects_listed - distinct_partitions,
'missing_one': object_copies_found[2],
'missing_two': object_copies_found[1],
'missing_all': object_copies_found[0],
'pct_found': value,
'copies_found': copies_found,
'copies_expected': copies_expected}
if __name__ == '__main__':
patcher.monkey_patch()
hubs.get_hub().debug_exceptions = False
parser = OptionParser(usage='''
Usage: %prog [options] [conf_file]
[conf_file] defaults to /etc/swift/stats.conf'''.strip())
parser.add_option('-j', '--dump-json', action='store_true', default=False,
help='dump dispersion report in json format')
options, args = parser.parse_args()
conffile = '/etc/swift/dispersion.conf'
if len(argv) == 2:
conffile = argv[1]
elif len(argv) > 2:
exit('Syntax: %s [conffile]' % argv[0])
if args:
conffile = args.pop(0)
c = ConfigParser()
if not c.read(conffile):
exit('Unable to read config file: %s' % conffile)
@ -241,6 +275,8 @@ if __name__ == '__main__':
dispersion_coverage = int(conf.get('dispersion_coverage', 1))
retries = int(conf.get('retries', 5))
concurrency = int(conf.get('concurrency', 25))
if options.dump_json or conf.get('dump_json', 'no').lower() in TRUE_VALUES:
json_output = True
coropool = GreenPool(size=concurrency)
@ -256,6 +292,11 @@ if __name__ == '__main__':
container_ring = Ring(os.path.join(swift_dir, 'container.ring.gz'))
object_ring = Ring(os.path.join(swift_dir, 'object.ring.gz'))
container_dispersion_report(coropool, connpool, account, container_ring,
retries)
object_dispersion_report(coropool, connpool, account, object_ring, retries)
container_result = container_dispersion_report(coropool, connpool,
account, container_ring,
retries)
object_result = object_dispersion_report(coropool, connpool, account,
object_ring, retries)
if json_output:
print json.dumps({"container": container_result,
"object": object_result})

View File

@ -24,7 +24,7 @@
.SH SYNOPSIS
.LP
.B swift-dispersion-report
.B swift-dispersion-report [-j|--dump-json] [conf_file]
.SH DESCRIPTION
.PP
@ -54,6 +54,12 @@ same configuration file, /etc/swift/dispersion.conf . The account used by these
tool should be a dedicated account for the dispersion stats and also have admin
privileges.
.SH OPTIONS
.RS 0
.PD 1
.IP "\fB-j, --dump-json\fR"
output dispersion report in json format
.SH CONFIGURATION
.PD 0
Example \fI/etc/swift/dispersion.conf\fR:
@ -67,6 +73,7 @@ Example \fI/etc/swift/dispersion.conf\fR:
.IP "# dispersion_coverage = 1"
.IP "# retries = 5"
.IP "# concurrency = 25"
.IP "# dump_json = no"
.RE
.PD
.SH EXAMPLE

View File

@ -221,6 +221,12 @@ place and then rerun the dispersion report::
100.00% of object copies found (7857 of 7857)
Sample represents 1.00% of the object partition space
Alternatively, the dispersion report can also be output in json format. This
allows it to be more easily consumed by third party utilities::
$ swift-dispersion-report -j
{"object": {"retries:": 0, "missing_two": 0, "copies_found": 7863, "missing_one": 0, "copies_expected": 7863, "pct_found": 100.0, "overlapping": 0, "missing_all": 0}, "container": {"retries:": 0, "missing_two": 0, "copies_found": 12534, "missing_one": 0, "copies_expected": 12534, "pct_found": 100.0, "overlapping": 15, "missing_all": 0}}
--------------------------------
Cluster Telemetry and Monitoring

View File

@ -6,3 +6,4 @@ auth_key = testing
# dispersion_coverage = 1
# retries = 5
# concurrency = 25
# dump_json = no