From: David-Sarah Hopwood Date: Fri, 22 Feb 2013 03:59:48 +0000 (+0000) Subject: leasedb/accounting crawler: only treat stable shares as disappeared or unleased. X-Git-Url: https://git.rkrishnan.org/something?a=commitdiff_plain;h=9f6d12691e511b112396a041091da2baf95ae18f;p=tahoe-lafs%2Ftahoe-lafs.git leasedb/accounting crawler: only treat stable shares as disappeared or unleased. fixes #1921 Signed-off-by: David-Sarah Hopwood --- diff --git a/docs/proposed/leasedb.rst b/docs/proposed/leasedb.rst index c0516c31..cab0e103 100644 --- a/docs/proposed/leasedb.rst +++ b/docs/proposed/leasedb.rst @@ -113,9 +113,9 @@ The accounting crawler may perform the following functions (but see ticket corrupted. This is handled in the same way as upgrading from a previous version. -- Detect shares that have unexpectedly disappeared from storage. The - disappearance of a share is logged, and its entry and leases are removed - from the leasedb. +- Detect shares with stable entries in the leasedb that have unexpectedly + disappeared from storage. The disappearance of a share is logged, and its + entry and leases are removed from the leasedb. Accounts diff --git a/src/allmydata/storage/accounting_crawler.py b/src/allmydata/storage/accounting_crawler.py index d54f6133..dd135b47 100644 --- a/src/allmydata/storage/accounting_crawler.py +++ b/src/allmydata/storage/accounting_crawler.py @@ -4,10 +4,12 @@ import time from twisted.internet import defer from allmydata.util.deferredutil import for_items +from allmydata.util.assertutil import _assert from allmydata.util import log from allmydata.storage.crawler import ShareCrawler from allmydata.storage.common import si_a2b -from allmydata.storage.leasedb import SHARETYPES, SHARETYPE_UNKNOWN, SHARETYPE_CORRUPTED +from allmydata.storage.leasedb import SHARETYPES, SHARETYPE_UNKNOWN, SHARETYPE_CORRUPTED, \ + STATE_STABLE class AccountingCrawler(ShareCrawler): @@ -73,7 +75,7 @@ class AccountingCrawler(ShareCrawler): # updated, and removed. for key, value in db_sharemap.iteritems(): (si_s, shnum) = key - (used_space, sharetype) = value + (used_space, sharetype, state) = value examined_sharesets[sharetype].add(si_s) @@ -92,30 +94,35 @@ class AccountingCrawler(ShareCrawler): stored_shares = set(stored_sharemap) db_shares = set(db_sharemap) - # add new shares to the DB + # Add new shares to the DB. new_shares = stored_shares - db_shares for shareid in new_shares: (si_s, shnum) = shareid - (used_space, sharetype) = stored_sharemap[shareid] + (used_space, sharetype, state) = stored_sharemap[shareid] self._leasedb.add_new_share(si_a2b(si_s), shnum, used_space, sharetype) self._leasedb.add_starter_lease(si_s, shnum) - # remove disappeared shares from DB - disappeared_shares = db_shares - stored_shares - for (si_s, shnum) in disappeared_shares: - log.msg(format="share SI=%(si_s)s shnum=%(shnum)s unexpectedly disappeared", - si_s=si_s, shnum=shnum, level=log.WEIRD) - # This is temporarily disabled, because it results in failures if we're examining - # a prefix while a share is created in it (ticket #1921). - #self._leasedb.remove_deleted_share(si_a2b(si_s), shnum) + # Remove disappeared shares from the DB. Note that only shares in STATE_STABLE + # should be considered "disappeared", since otherwise it would be possible for + # this to delete shares that are in the process of being created (see ticket #1921). + potentially_disappeared_shares = db_shares - stored_shares + for shareid in potentially_disappeared_shares: + (used_space, sharetype, state) = db_sharemap[shareid] + if state == STATE_STABLE: + (si_s, shnum) = shareid + log.msg(format="share SI=%(si_s)s shnum=%(shnum)s unexpectedly disappeared", + si_s=si_s, shnum=shnum, level=log.WEIRD) + self._leasedb.remove_deleted_share(si_a2b(si_s), shnum) recovered_sharesets = [set() for st in xrange(len(SHARETYPES))] def _delete_share(ign, key, value): (si_s, shnum) = key - (used_space, sharetype) = value + (used_space, sharetype, state) = value + _assert(state == STATE_STABLE, state=state) storage_index = si_a2b(si_s) + d3 = defer.succeed(None) def _mark_and_delete(ign): self._leasedb.mark_share_as_going(storage_index, shnum) @@ -141,6 +148,7 @@ class AccountingCrawler(ShareCrawler): d3.addCallbacks(_deleted, _not_deleted) return d3 + # This only includes stable unleased shares (see ticket #1921). unleased_sharemap = self._leasedb.get_unleased_shares_for_prefix(prefix) d2 = for_items(_delete_share, unleased_sharemap) diff --git a/src/allmydata/storage/leasedb.py b/src/allmydata/storage/leasedb.py index dbd72f90..02524e78 100644 --- a/src/allmydata/storage/leasedb.py +++ b/src/allmydata/storage/leasedb.py @@ -137,14 +137,15 @@ class LeaseDB: def get_shares_for_prefix(self, prefix): """ - Returns a dict mapping (si_s, shnum) pairs to (used_space, sharetype) pairs. + Returns a dict mapping (si_s, shnum) pairs to (used_space, sharetype, state) triples + for shares with this prefix. """ - self._cursor.execute("SELECT `storage_index`,`shnum`, `used_space`, `sharetype`" + self._cursor.execute("SELECT `storage_index`,`shnum`, `used_space`, `sharetype`, `state`" " FROM `shares`" " WHERE `prefix` == ?", (prefix,)) - db_sharemap = dict([((str(si_s), int(shnum)), (int(used_space), int(sharetype))) - for (si_s, shnum, used_space, sharetype) in self._cursor.fetchall()]) + db_sharemap = dict([((str(si_s), int(shnum)), (int(used_space), int(sharetype), int(state))) + for (si_s, shnum, used_space, sharetype, state) in self._cursor.fetchall()]) return db_sharemap def add_new_share(self, storage_index, shnum, used_space, sharetype): @@ -284,18 +285,24 @@ class LeaseDB: return map(_to_age, rows) def get_unleased_shares_for_prefix(self, prefix): + """ + Returns a dict mapping (si_s, shnum) pairs to (used_space, sharetype, state) triples + for stable, unleased shares with this prefix. + """ if self.debug: print "GET_UNLEASED_SHARES_FOR_PREFIX", prefix # This would be simpler, but it doesn't work because 'NOT IN' doesn't support multiple columns. - #query = ("SELECT `storage_index`, `shnum`, `used_space`, `sharetype` FROM `shares`" - # " WHERE (`storage_index`, `shnum`) NOT IN (SELECT DISTINCT `storage_index`, `shnum` FROM `leases`)") + #query = ("SELECT `storage_index`, `shnum`, `used_space`, `sharetype`, `state` FROM `shares`" + # " WHERE `state` = STATE_STABLE " + # " AND (`storage_index`, `shnum`) NOT IN (SELECT DISTINCT `storage_index`, `shnum` FROM `leases`)") # This "negative join" should be equivalent. - self._cursor.execute("SELECT DISTINCT s.storage_index, s.shnum, s.used_space, s.sharetype FROM `shares` s LEFT JOIN `leases` l" + self._cursor.execute("SELECT DISTINCT s.storage_index, s.shnum, s.used_space, s.sharetype, s.state" + " FROM `shares` s LEFT JOIN `leases` l" " ON (s.storage_index = l.storage_index AND s.shnum = l.shnum)" - " WHERE s.prefix = ? AND l.storage_index IS NULL", - (prefix,)) - db_sharemap = dict([((str(si_s), int(shnum)), (int(used_space), int(sharetype))) - for (si_s, shnum, used_space, sharetype) in self._cursor.fetchall()]) + " WHERE s.prefix = ? AND s.state = ? AND l.storage_index IS NULL", + (prefix, STATE_STABLE)) + db_sharemap = dict([((str(si_s), int(shnum)), (int(used_space), int(sharetype), int(state))) + for (si_s, shnum, used_space, sharetype, state) in self._cursor.fetchall()]) return db_sharemap def remove_leases_by_renewal_time(self, renewal_cutoff_time):