bytemark/flexnbd-c
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Simplify the migration handover protocol

Browse Source 
The three-way hand-off has a problem: there's no way to arrange for the
state of the migration to be unambiguous in case of failure.  If the
final "disconnect" message is lost (as in, the destination never
receives it whether it is sent by the sender or not), the destination
has no option but to quit with an error status and let a human sort it
out.  However, at that point we can either arrange to have a .INCOMPLETE
file still on disc or not - and it doesn't matter which we choose, we
can still end up with dataloss by picking a specific calamity to have
befallen the sender.

Given this, it makes sense to fall back to a simpler protocol: just send
all the data, then send a "disconnect" message.  This has the same
downside that we need a human to sort out specific failure cases, but
combined with --unlink before sending "disconnect" (see next patch) it
will always be possible for a human to disambiguate, whether the
destination quit with an error status or not.
This commit is contained in:
Alex Young
2012-07-23 10:22:25 +01:00
parent f6f4266fd6
commit fd935ce4c9
16 changed files with 54 additions and 212 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
tests/acceptance/fakes/dest/close_after_entrust.rb
View File

@@ -1,29 +0,0 @@
#!/usr/bin/env ruby
# encoding: utf-8
# Open a server, accept a client, then we expect a single write
# followed by an entrust. Disconnect after the entrust. We expect a
# reconnection followed by a full mirror.
require 'flexnbd/fake_dest'
include FlexNBD
addr, port, src_pid = *ARGV
server = FakeDest.new( addr, port )
client = server.accept
client.write_hello
write_req = client.read_request
data = client.read_data( write_req[:len] )
client.write_reply( write_req[:handle], 0 )
entrust_req = client.read_request
fail "Not an entrust" unless entrust_req[:type] == 65536
client.close
client2 = server.accept
client2.receive_mirror
exit(0)

6
tests/acceptance/fakes/dest/close_after_writes.rb
View File

@@ -3,7 +3,8 @@
# Open a server, accept a client, then we expect a single write
# followed by an entrust. However, we disconnect after the write so
# the entrust will fail. We expect a reconnection.
# the entrust will fail. We don't expect a reconnection: the sender
# can't reliably spot a failed send.
require 'flexnbd/fake_dest'
include FlexNBD
@@ -21,7 +22,4 @@ client.write_reply( req[:handle], 0 )
client.close
Process.kill("CONT", src_pid.to_i)
client2 = server.accept
client2.close
exit(0)

34
tests/acceptance/fakes/dest/error_on_entrust.rb
View File

@@ -1,34 +0,0 @@
#!/usr/bin/env ruby
# encoding: utf-8
# Receive a mirror, but respond to the entrust with an error. There's
# currently no code path in flexnbd which can do this, but we could
# add one.
require 'flexnbd/fake_dest'
include FlexNBD
addr, port = *ARGV
server = FakeDest.new( addr, port )
client = server.accept
client.write_hello
loop do
req = client.read_request
if req[:type] == 1
client.read_data( req[:len] )
client.write_reply( req[:handle] )
else
client.write_reply( req[:handle], 1 )
break
end
end
client.close
client2 = server.accept( "Timed out waiting for a reconnection" )
client2.close
server.close
exit(0)