projects / ipfire-2.x.git / blame
| Commit | Line | Data |
|---|---|---|
| 2cb7cef9 BS |
1 | From: NeilBrown <neilb@suse.de> |
| 2 | Subject: reset TCP connect timeout on successful connection. | |
| 3 | Patch-mainline: no - see below | |
| 4 | References: bnc#498708 | |
| 5 | ||
| 6 | Currently, every time an NFS server closes a TCP connection, the | |
| 7 | client will double its time out (up to 300 seconds) before attempting | |
| 8 | to reconnect. | |
| 9 | The timeout should only double while it fails. On success it should | |
| 10 | reset the timeout. | |
| 11 | ||
| 12 | This patch resets the timeout on a successful connection. | |
| 13 | It also ensures that we the timeout it doubled, it does not remain at | |
| 14 | 0. This should fix a behaviour which has been observed where the | |
| 15 | client hammers on the server until it accepts a connection. | |
| 16 | ||
| 17 | Note: this has not been accepted upstream. | |
| 18 | The stated reason is that some (Netapp?) servers are prone to accept | |
| 19 | a connection and then drop it immediately. This would defeat the backoff. | |
| 20 | While this should be fixed there are complexities in getting it right, | |
| 21 | and the current patch is at least an improvement. | |
| 22 | (there where already cases where the client would retry at a high | |
| 23 | rate and the new case is, I think, quite remote). | |
| 24 | When things are sorted out with upstream, we can get a better fix. | |
| 25 | ||
| 26 | Signed-off-by: NeilBrown <neilb@suse.de> | |
| 27 | ||
| 28 | --- | |
| 29 | net/sunrpc/xprtsock.c | 3 +++ | |
| 30 | 1 file changed, 3 insertions(+) | |
| 31 | ||
| 32 | --- a/net/sunrpc/xprtsock.c | |
| 33 | +++ b/net/sunrpc/xprtsock.c | |
| 34 | @@ -1143,6 +1143,7 @@ static void xs_tcp_state_change(struct s | |
| 35 | TCP_RCV_COPY_FRAGHDR | TCP_RCV_COPY_XID; | |
| 36 | ||
| 37 | xprt_wake_pending_tasks(xprt, 0); | |
| 38 | + xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO; | |
| 39 | } | |
| 40 | spin_unlock_bh(&xprt->transport_lock); | |
| 41 | break; | |
| 42 | @@ -1785,6 +1786,8 @@ static void xs_connect(struct rpc_task * | |
| 43 | &transport->connect_worker, | |
| 44 | xprt->reestablish_timeout); | |
| 45 | xprt->reestablish_timeout <<= 1; | |
| 46 | + if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO) | |
| 47 | + xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO; | |
| 48 | if (xprt->reestablish_timeout > XS_TCP_MAX_REEST_TO) | |
| 49 | xprt->reestablish_timeout = XS_TCP_MAX_REEST_TO; | |
| 50 | } else { |