[PATCH] Fix bug 9412 - SMB2 server doesn't support recvfile.
jones.kstw at gmail.com
Thu Apr 17 01:19:33 MDT 2014
Excuse me to resend this mail,
looks like the size of attached file is too big,
hence put it to alternative link:
2014-04-17 13:15 GMT+08:00 Jones <jones.kstw at gmail.com>:
> 2014-04-17 5:34 GMT+08:00 Jeremy Allison <jra at samba.org>:
> Can you give me more details as to
>> how you're reproducing this problem ?
> Hello Jeremy,
> The attached picture is what I reproduce the EAGAIN,
> hmm hope following description is not too chatter.
> Over a normal LAN which owned RTT is always <1ms,
> cannot easily see splice returned -1 with EAGAIN.
> Hence try to increase the RTT through the WANem,
> and manually do something to make EAGAIN pop-out.
> A) Environment setup
> The network topology is like this:
> Client: Windows 7
> Server: Linux box as a NAS
> Latency generator: PC with WANem installed and located between client &
> The latency generator is came from WANem ISO file,
> the WANem v3.0 Beta released 20130207 is selected,
> a bootable live CD is installed into the PC.
> PC's eth1 is directly linked to the client: windows 7,
> PC's eth2 is directly linked to the server: my linux box.
> The ip address arrangement is as attached picture,
> Client: 172.16.1.100/24
> PCeth1: 172.16.1.101/24
> PCeth2: 172.16.2.201/24
> Server: 172.16.2.200/24
> Then assign the routing rules under client & server.
> For Client (Windows 7):
> Start -> Run -> cmd.exe,
> input following commands:
> route add 172.16.2.200 mask 255.255.255.255 172.16.1.101
> route add 172.16.2.201 mask 255.255.255.255 172.16.1.101
> For Server (Linux box as a NAS):
> input following commands:
> route add -host 172.16.2.200 netmask 0.0.0.0 gw 172.16.2.201; \
> route add -host 172.16.1.100 netmask 0.0.0.0 gw 172.16.2.201; \
> route add -host 172.16.1.101 netmask 0.0.0.0 gw 172.16.2.201;
> B) Preliminary ping
> Go to WANem GUI -> Basic Mode -> increase eth2 delay time -> Apply
> Then on client site try ping the ip address of server,
> and check the output message if RTT field is raised as expected.
> If yes, this environment setup is good to go!
> And click Reset settings button to rollback as default now.
> C) Start this test case
> Since the client is windows 7,
> hence the remote shared folder resident in linux box is
> mapped to client site as a net drive with letter -> y: .
> Drag-and-drop a single 8GB file from local disk c: to net disk y: ,
> during uploading is in progress,
> launch the strace as following:
> strace 2>&1 -p $smbd_pid | grep -inr --color -E "(splice|fcntl|pwrite)"
> and a bulk of splice calls are prompted.
> Next is through the WANem GUI and increase the eth2 delay time to 500ms,
> and click the Apply settings button.
> Now the upload progress and splice prompt rate would be slow down,
> on this phase the EAGAIN still not happened.
> And last ... but not the least!
> Manually click the Apply settings button triple times and wait half a
> and loop this movement,
> Then the EAGAIN is in action with luck!
> Hope this help reproduce the EAGAIN,
> thank you.
More information about the samba-technical