RepoMirrors/ceph
1
0
Fork 0
mirror of https://github.com/ceph/ceph synced 2025-02-20 17:37:29 +00:00
Code Issues Packages Projects Releases Wiki Activity

tasks/resolve_stuck_peering: adjust to new ctx.manager location

Browse Source 
This task was re-added after the multi-cluster branch merged.

Signed-off-by: Josh Durgin <jdurgin@redhat.com>
This commit is contained in:
Josh Durgin 2016-05-11 12:01:52 -07:00
parent 0cade258c8
commit ff942710ba
1 changed files with 18 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
tasks/resolve_stuck_peering.py
View File

@ -20,11 +20,13 @@ def task(ctx, config):
assert isinstance(config, dict), \
'Resolve stuck peering only accepts a dict for config'
while len(ctx.manager.get_osd_status()['up']) < 3:
manager = ctx.managers['ceph']
while len(manager.get_osd_status()['up']) < 3:
time.sleep(10)
ctx.manager.wait_for_clean()
manager.wait_for_clean()
dummyfile = '/etc/fstab'
dummyfile1 = '/etc/resolv.conf'
@ -32,16 +34,16 @@ def task(ctx, config):
#create 1 PG pool
pool='foo'
log.info('creating pool foo')
ctx.manager.raw_cluster_cmd('osd', 'pool', 'create', '%s' % pool, '1')
manager.raw_cluster_cmd('osd', 'pool', 'create', '%s' % pool, '1')
#set min_size of the pool to 1
#so that we can continue with I/O
#when 2 osds are down
ctx.manager.set_pool_property(pool, "min_size", 1)
manager.set_pool_property(pool, "min_size", 1)
osds = [0, 1, 2]
primary = ctx.manager.get_pg_primary('foo', 0)
primary = manager.get_pg_primary('foo', 0)
log.info("primary osd is %d", primary)
others = list(osds)
@ -54,32 +56,32 @@ def task(ctx, config):
for i in range(100):
rados(ctx, mon, ['-p', 'foo', 'put', 'existing_%d' % i, dummyfile])
ctx.manager.wait_for_clean()
manager.wait_for_clean()
#kill other osds except primary
log.info('killing other osds except primary')
for i in others:
ctx.manager.kill_osd(i)
manager.kill_osd(i)
for i in others:
ctx.manager.mark_down_osd(i)
manager.mark_down_osd(i)
for i in range(100):
rados(ctx, mon, ['-p', 'foo', 'put', 'new_%d' % i, dummyfile1])
#kill primary osd
ctx.manager.kill_osd(primary)
ctx.manager.mark_down_osd(primary)
manager.kill_osd(primary)
manager.mark_down_osd(primary)
#revive other 2 osds
for i in others:
ctx.manager.revive_osd(i)
manager.revive_osd(i)
#make sure that pg is down
#Assuming pg number for single pg pool will start from 0
pgnum=0
pgstr = ctx.manager.get_pgid(pool, pgnum)
stats = ctx.manager.get_single_pg_stats(pgstr)
pgstr = manager.get_pgid(pool, pgnum)
stats = manager.get_single_pg_stats(pgstr)
print stats['state']
timeout=60
@ -90,19 +92,19 @@ def task(ctx, config):
'failed to reach down state before timeout expired'
#mark primary as lost
ctx.manager.raw_cluster_cmd('osd', 'lost', '%d' % primary,\
manager.raw_cluster_cmd('osd', 'lost', '%d' % primary,\
'--yes-i-really-mean-it')
#expect the pg status to be active+undersized+degraded
#pg should recover and become active+clean within timeout
stats = ctx.manager.get_single_pg_stats(pgstr)
stats = manager.get_single_pg_stats(pgstr)
print stats['state']
timeout=10
start=time.time()
while ctx.manager.get_num_down():
while manager.get_num_down():
assert time.time() - start < timeout, \
'failed to recover before timeout expired'