git-annex/Assistant/TransferrerPool.hs
Joey Hess e5b4d447b6 assistant: Start a new git-annex transferkeys process after a network connection change
So that remotes that use a persistent network connection are restarted.

A remote might keep open a long duration network connection, and could
fail to deal well with losing the connection. This is particularly a
concern now that we have external special reotes. An external
special remote that is implemented naively might open the connection only
when PREPARE is sent, and if it loses connection, throw errors on each
request that is made.

(Note that the ssh connection caching should not have this problem; if the
long-duration ssh process loses connection, the named pipe is disconnected
and the next ssh attempt will reconnect. Also, XMPP already deals with
disconnection robustly in its own way.)

There's no way for git-annex to know if a lost network connection actually
affects a given remote, which might have a transfer in process. It does not
make sense to force kill the transferkeys process every time the NetWatcher
detects a change. (Especially because the NetWatcher sometimes polls 1
change per hour.)

In any case, the NetWatcher only detects connection to a network, not
disconnection. So if a transfer is in progress over the network, and the
network goes down, that will need to time out on its own.

An alternate approch that was considered is to use a separate transferkeys
process for each remote, and detect when a request fails, and assume that
means that process is in a failing state and restart it. The problem with
that approach is that if a resource is not available and a remote fails
every time, it degrades to starting a new transferkeys process for every
file transfer, which is too expensive.

Instead, this commit only handles the network reconnection case, and restarts
transferkeys only once the network has reconnected and another transfer needs
to be made. So, a transferkeys process will be reused for 1 hour, or until the
next network connection.

----

The NotificationBroadcaster was rewritten to use TMVars rather than MSampleVars,
to allow checking without blocking if a notification has been received.

----

This commit was sponsored by Tobias Brunner.
2014-01-06 16:03:39 -04:00

95 lines
3.2 KiB
Haskell

{- A pool of "git-annex transferkeys" processes
-
- Copyright 2013 Joey Hess <joey@kitenet.net>
-
- Licensed under the GNU GPL version 3 or higher.
-}
module Assistant.TransferrerPool where
import Assistant.Common
import Assistant.Types.TransferrerPool
import Logs.Transfer
import Utility.Batch
import qualified Command.TransferKeys as T
import Control.Concurrent.STM hiding (check)
import System.Process (create_group, std_in, std_out)
import Control.Exception (throw)
import Control.Concurrent
{- Runs an action with a Transferrer from the pool.
-
- Only one Transferrer is left running in the pool at a time.
- So if this needed to start a new Transferrer, it's stopped when done.
-}
withTransferrer :: FilePath -> BatchCommandMaker -> TransferrerPool -> (Transferrer -> IO a) -> IO a
withTransferrer program batchmaker pool a = do
i@(TransferrerPoolItem (Just t) _) <- maybe
(mkTransferrerPoolItem pool =<< mkTransferrer program batchmaker)
(checkTransferrerPoolItem program batchmaker)
=<< atomically (popTransferrerPool pool)
v <- tryNonAsync $ a t
sz <- atomically $ pushTransferrerPool pool i
when (sz > 1) $
void $ forkIO $ stopTransferrer t
either throw return v
{- Check if a Transferrer from the pool is still ok to be used.
- If not, stop it and start a new one. -}
checkTransferrerPoolItem :: FilePath -> BatchCommandMaker -> TransferrerPoolItem -> IO TransferrerPoolItem
checkTransferrerPoolItem program batchmaker i = case i of
TransferrerPoolItem (Just t) check -> ifM check
( return i
, do
stopTransferrer t
new check
)
TransferrerPoolItem Nothing check -> new check
where
new check = do
t <- mkTransferrer program batchmaker
return $ TransferrerPoolItem (Just t) check
{- Requests that a Transferrer perform a Transfer, and waits for it to
- finish. -}
performTransfer :: Transferrer -> Transfer -> AssociatedFile -> IO Bool
performTransfer transferrer t f = catchBoolIO $ do
T.sendRequest t f (transferrerWrite transferrer)
T.readResponse (transferrerRead transferrer)
{- Starts a new git-annex transferkeys process, setting up handles
- that will be used to communicate with it. -}
mkTransferrer :: FilePath -> BatchCommandMaker -> IO Transferrer
mkTransferrer program batchmaker = do
{- It runs as a batch job. -}
let (program', params') = batchmaker (program, [Param "transferkeys"])
{- It's put into its own group so that the whole group can be
- killed to stop a transfer. -}
(Just writeh, Just readh, _, pid) <- createProcess
(proc program' $ toCommand params')
{ create_group = True
, std_in = CreatePipe
, std_out = CreatePipe
}
fileEncoding readh
fileEncoding writeh
return $ Transferrer
{ transferrerRead = readh
, transferrerWrite = writeh
, transferrerHandle = pid
}
{- Checks if a Transferrer is still running. If not, makes a new one. -}
checkTransferrer :: FilePath -> BatchCommandMaker -> Transferrer -> IO Transferrer
checkTransferrer program batchmaker t =
maybe (return t) (const $ mkTransferrer program batchmaker)
=<< getProcessExitCode (transferrerHandle t)
{- Closing the fds will stop the transferrer. -}
stopTransferrer :: Transferrer -> IO ()
stopTransferrer t = do
hClose $ transferrerRead t
hClose $ transferrerWrite t
void $ waitForProcess $ transferrerHandle t