3 I contain the client-side code which speaks to storage servers, in particular
4 the foolscap-based server implemented in src/allmydata/storage/*.py .
9 # 1: implement StorageFarmBroker (i.e. "storage broker"), change Client to
10 # create it, change uploader/servermap to get rrefs from it. ServerFarm calls
11 # IntroducerClient.subscribe_to . ServerFarm hides descriptors, passes rrefs
12 # to clients. webapi status pages call broker.get_info_about_serverid.
14 # 2: move get_info methods to the descriptor, webapi status pages call
15 # broker.get_descriptor_for_serverid().get_info
17 # 3?later?: store descriptors in UploadResults/etc instead of serverids,
18 # webapi status pages call descriptor.get_info and don't use storage_broker
21 # 4: enable static config: tahoe.cfg can add descriptors. Make the introducer
22 # optional. This closes #467
24 # 5: implement NativeStorageClient, pass it to Tahoe2PeerSelector and other
25 # clients. Clients stop doing callRemote(), use NativeStorageClient methods
26 # instead (which might do something else, i.e. http or whatever). The
27 # introducer and tahoe.cfg only create NativeStorageClients for now.
29 # 6: implement other sorts of IStorageClient classes: S3, etc
33 from zope.interface import implements, Interface
34 from foolscap.api import eventually
35 from allmydata.interfaces import IStorageBroker
36 from allmydata.util import idlib, log
37 from allmydata.util.assertutil import precondition
38 from allmydata.util.rrefutil import add_version_to_remote_reference
39 from allmydata.util.hashutil import sha1
41 # who is responsible for de-duplication?
43 # IC remembers the unpacked announcements it receives, to provide for late
44 # subscribers and to remove duplicates
46 # if a client subscribes after startup, will they receive old announcements?
49 # who will be responsible for signature checking?
50 # make it be IntroducerClient, so they can push the filter outwards and
51 # reduce inbound network traffic
53 # what should the interface between StorageFarmBroker and IntroducerClient
55 # don't pass signatures: only pass validated blessed-objects
57 class StorageFarmBroker:
58 implements(IStorageBroker)
59 """I live on the client, and know about storage servers. For each server
60 that is participating in a grid, I either maintain a connection to it or
61 remember enough information to establish a connection to it on demand.
62 I'm also responsible for subscribing to the IntroducerClient to find out
63 about new servers as they are announced by the Introducer.
65 def __init__(self, tub, permute_peers):
67 assert permute_peers # False not implemented yet
68 self.permute_peers = permute_peers
69 # self.servers maps serverid -> IServer, and keeps track of all the
70 # storage servers that we've heard about. Each descriptor manages its
71 # own Reconnector, and will give us a RemoteReference when we ask
74 self.introducer_client = None
76 # these two are used in unit tests
77 def test_add_rref(self, serverid, rref):
78 s = NativeStorageServer(serverid, {})
80 self.servers[serverid] = s
82 def test_add_server(self, serverid, s):
83 self.servers[serverid] = s
85 def use_introducer(self, introducer_client):
86 self.introducer_client = ic = introducer_client
87 ic.subscribe_to("storage", self._got_announcement)
89 def _got_announcement(self, serverid, ann_d):
90 precondition(isinstance(serverid, str), serverid)
91 precondition(len(serverid) == 20, serverid)
92 assert ann_d["service-name"] == "storage"
93 old = self.servers.get(serverid)
95 if old.get_announcement() == ann_d:
98 del self.servers[serverid]
100 # now we forget about them and start using the new one
101 dsc = NativeStorageServer(serverid, ann_d)
102 self.servers[serverid] = dsc
103 dsc.start_connecting(self.tub, self._trigger_connections)
104 # the descriptor will manage their own Reconnector, and each time we
105 # need servers, we'll ask them if they're connected or not.
107 def _trigger_connections(self):
108 # when one connection is established, reset the timers on all others,
109 # to trigger a reconnection attempt in one second. This is intended
110 # to accelerate server connections when we've been offline for a
111 # while. The goal is to avoid hanging out for a long time with
112 # connections to only a subset of the servers, which would increase
113 # the chances that we'll put shares in weird places (and not update
114 # existing shares of mutable files). See #374 for more details.
115 for dsc in self.servers.values():
118 def get_servers_for_psi(self, peer_selection_index):
119 # return a list of server objects (IServers)
120 assert self.permute_peers == True
121 def _permuted(server):
122 seed = server.get_permutation_seed()
123 return sha1(peer_selection_index + seed).digest()
124 return sorted(self.get_connected_servers(), key=_permuted)
126 def get_all_serverids(self):
127 return frozenset(self.servers.keys())
129 def get_connected_servers(self):
130 return frozenset([s for s in self.servers.values() if s.get_rref()])
132 def get_known_servers(self):
133 return frozenset(self.servers.values())
135 def get_nickname_for_serverid(self, serverid):
136 if serverid in self.servers:
137 return self.servers[serverid].get_nickname()
141 class IServer(Interface):
142 """I live in the client, and represent a single server."""
143 def start_connecting(tub, trigger_cb):
150 class NativeStorageServer:
151 """I hold information about a storage server that we want to connect to.
152 If we are connected, I hold the RemoteReference, their host address, and
153 the their version information. I remember information about when we were
154 last connected too, even if we aren't currently connected.
156 @ivar announcement_time: when we first heard about this service
157 @ivar last_connect_time: when we last established a connection
158 @ivar last_loss_time: when we last lost a connection
160 @ivar version: the server's versiondict, from the most recent announcement
161 @ivar nickname: the server's self-reported nickname (unicode), same
163 @ivar rref: the RemoteReference, if connected, otherwise None
164 @ivar remote_host: the IAddress, if connected, otherwise None
169 "http://allmydata.org/tahoe/protocols/storage/v1" :
170 { "maximum-immutable-share-size": 2**32,
171 "tolerates-immutable-read-overrun": False,
172 "delete-mutable-shares-with-zero-length-writev": False,
174 "application-version": "unknown: no get_version()",
177 def __init__(self, serverid, ann_d, min_shares=1):
178 self.serverid = serverid
179 self._tubid = serverid
180 self.announcement = ann_d
181 self.min_shares = min_shares
183 self.serverid_s = idlib.shortnodeid_b2a(self.serverid)
184 self.announcement_time = time.time()
185 self.last_connect_time = None
186 self.last_loss_time = None
187 self.remote_host = None
189 self._reconnector = None
190 self._trigger_cb = None
193 return "<NativeStorageServer for %s>" % self.get_name()
194 def get_serverid(self):
196 def get_permutation_seed(self):
198 def get_version(self):
200 return self.rref.version
202 def get_name(self): # keep methodname short
203 return self.serverid_s
204 def get_longname(self):
205 return idlib.nodeid_b2a(self._tubid)
206 def get_lease_seed(self):
208 def get_foolscap_write_enabler_seed(self):
211 def get_nickname(self):
212 return self.announcement["nickname"].decode("utf-8")
213 def get_announcement(self):
214 return self.announcement
215 def get_remote_host(self):
216 return self.remote_host
217 def get_last_connect_time(self):
218 return self.last_connect_time
219 def get_last_loss_time(self):
220 return self.last_loss_time
221 def get_announcement_time(self):
222 return self.announcement_time
224 def start_connecting(self, tub, trigger_cb):
225 furl = self.announcement["FURL"]
226 self._trigger_cb = trigger_cb
227 self._reconnector = tub.connectTo(furl, self._got_connection)
229 def _got_connection(self, rref):
230 lp = log.msg(format="got connection to %(name)s, getting versions",
231 name=self.get_name(),
232 facility="tahoe.storage_broker", umid="coUECQ")
234 eventually(self._trigger_cb)
235 default = self.VERSION_DEFAULTS
236 d = add_version_to_remote_reference(rref, default)
237 d.addCallback(self._got_versioned_service, lp)
238 d.addErrback(log.err, format="storageclient._got_connection",
239 name=self.get_name(), umid="Sdq3pg")
241 def _got_versioned_service(self, rref, lp):
242 log.msg(format="%(name)s provided version info %(version)s",
243 name=self.get_name(), version=rref.version,
244 facility="tahoe.storage_broker", umid="SWmJYg",
245 level=log.NOISY, parent=lp)
247 self.last_connect_time = time.time()
248 self.remote_host = rref.getPeer()
250 rref.notifyOnDisconnect(self._lost)
256 log.msg(format="lost connection to %(name)s", name=self.get_name(),
257 facility="tahoe.storage_broker", umid="zbRllw")
258 self.last_loss_time = time.time()
260 self.remote_host = None
262 def stop_connecting(self):
263 # used when this descriptor has been superceded by another
264 self._reconnector.stopConnecting()
266 def try_to_connect(self):
267 # used when the broker wants us to hurry up
268 self._reconnector.reset()
270 class UnknownServerTypeError(Exception):