Hi all,

We have come across a gridftp problem that arises when you do an lcg-cp.
At our site we have two storage elements. One is a RH7.3 machine and the
other is an IRIX machine. This problem occurs when the file is on the
IRIX machine. Sometimes lcg-cp works and sometimes it doesn't. When it
doesn't work I get the following output:

[ron@mu7 ron]$ lcg-cp -v --vo dteam lfn:aap file:///tmp/pipoaap3
Source URL: lfn:aap
File size: 114
Source URL for copy:
gsiftp://teras.sara.nl/home/dteam/generated/2004-08-30/file
3be54fc5-4fd7-4cde-97ba-fb8a18afae14
Destination URL: file:///tmp/pipoaap3
# streams: 1
a system call failed (Connection refused)

In the SYSLOG (the IRIX equivalent of /var/log/messages) I get the
following when it doesn't work:
Aug 31 13:31:32 6D:p1 gridftpd[2319522]: GSSAPI user
/O=dutchgrid/O=users/O=sara
/CN=Ron Trompert is authorized as dteam
Aug 31 13:31:32 6D:p1 gridftpd[2319522]: USER :globus-mapping:
Aug 31 13:31:32 6D:p1 gridftpd[2319522]: PASS password
Aug 31 13:31:32 6D:p1 gridftpd[2319522]: FTP LOGIN FROM
mu7.matrix.sara.nl [145.
100.29.135], dteam
Aug 31 13:31:32 6D:p1 gridftpd[2319522]: FEAT
Aug 31 13:31:32 6D:p1 gridftpd[2319522]: TYPE Image
Aug 31 13:31:32 6D:p1 gridftpd[2319522]: SIZE
/home/dteam/generated/2004-08-31/f
ile2bc86bb3-d866-4836-974c-e25b6042f6d1
Aug 31 13:31:32 6D:p1 gridftpd[2319522]: QUIT

And when it works I get:
Aug 31 10:51:39 6D:p1 gridftpd[2247024]: GSSAPI user
/O=dutchgrid/O=users/O=sara
/CN=Ron Trompert is authorized as dteam
Aug 31 10:51:39 6D:p1 gridftpd[2247024]: USER :globus-mapping:
Aug 31 10:51:40 6D:p1 gridftpd[2247024]: PASS password
Aug 31 10:51:40 6D:p1 gridftpd[2247024]: FTP LOGIN FROM
mu7.matrix.sara.nl [145.
100.29.135], dteam
Aug 31 10:51:40 6D:p1 gridftpd[2247024]: FEAT
Aug 31 10:51:40 6D:p1 gridftpd[2247024]: TYPE Image
Aug 31 10:51:40 6D:p1 gridftpd[2247024]: SIZE
/home/dteam/generated/2004-08-30/f
ile3be54fc5-4fd7-4cde-97ba-fb8a18afae14
Aug 31 10:51:40 6D:p1 gridftpd[2247024]: DCAU
Aug 31 10:51:40 6D:p1 gridftpd[2247024]: PASV
Aug 31 10:51:40 6D:p1 gridftpd[2247024]: RETR
/home/dteam/generated/2004-08-30/f
ile3be54fc5-4fd7-4cde-97ba-fb8a18afae14
Aug 31 10:51:40 6D:p1 gridftpd[2247024]: QUIT

It looks like DCAU doesn't get through somehow. The snoop/tcpdump output
of the IRIX SE and UI (on which lcg-cp was run) is below. The gridftp
server is teras.sara.nl and the UI mu7.matrix.sara.nl.

On SE:
12:26:49.615028 teras.sara.nl -> mu7.matrix.sara.nl TCP D=20000 S=2811
Ack=2
503864890 Seq=902688248 Len=0 Win=64512 Options=<nop,nop,tstamp 1295625
42799191
>
12:26:49.621455 teras.sara.nl -> mu7.matrix.sara.nl TCP D=20000 S=2811
Ack=2
503864890 Seq=902688248 Len=110 Win=64512 Options=<nop,nop,tstamp
1295625 427991
91>
12:26:49.642044 teras.sara.nl -> mu7.matrix.sara.nl TCP D=20000 S=2811
Fin Ack=2
503864890 Seq=902688358 Len=338 Win=64512 Options=<nop,nop,tstamp
1295625 427991
91>
12:26:49.644000 mu7.matrix.sara.nl -> teras.sara.nl TCP D=2811 S=20000
Fin Ack=9
02688697 Seq=2503864890 Len=0 Win=20272 Options=<nop,nop,tstamp 42799194
1295625
>
12:26:49.644077 teras.sara.nl -> mu7.matrix.sara.nl TCP D=20000 S=2811
Ack=2
503864891 Seq=902688697 Len=0 Win=64512 Options=<nop,nop,tstamp 1295625
42799194
>
12:26:49.654552 mu7.matrix.sara.nl -> teras.sara.nl TCP D=2811 S=20000
Syn Seq=2
504656570 Len=0 Win=5840 Options=<mss 1460,sackOK,tstamp 42799195
0,nop,wscale 0
>
12:26:49.654604 teras.sara.nl -> mu7.matrix.sara.nl TCP D=20000 S=2811
Rst Ack=2
503864891 Seq=902688697 Len=0 Win=64512
12:26:49.654651 teras.sara.nl -> mu7.matrix.sara.nl TCP D=20000 S=2811
Rst Ack=2
504656571 Win=0

On UI:
12:26:49.603858 teras.sara.nl.2811 > mu7.matrix.sara.nl.20000: . ack
5640 win 64512 <nop,nop,timestamp 1295625 42799190> (DF) [tos 0x10]
12:26:49.612500 teras.sara.nl.2811 > mu7.matrix.sara.nl.20000: P
8051:8109(58) ack 5640 win 64512 <nop,nop,timestamp 1295625 42799190>
(DF) [tos 0x10]
12:26:49.612760 mu7.matrix.sara.nl.20000 > teras.sara.nl.2811: P
5640:5698(58) ack 8109 win 20272 <nop,nop,timestamp 42799191 1295625>
(DF)
12:26:49.614399 teras.sara.nl.2811 > mu7.matrix.sara.nl.20000: . ack
5698 win 64512 <nop,nop,timestamp 1295625 42799191> (DF) [tos 0x10]
12:26:49.620846 teras.sara.nl.2811 > mu7.matrix.sara.nl.20000: P
8109:8219(110) ack 5698 win 64512 <nop,nop,timestamp 1295625 42799191>
(DF) [tos 0x10]
12:26:49.641471 teras.sara.nl.2811 > mu7.matrix.sara.nl.20000: FP
8219:8557(338) ack 5698 win 64512 <nop,nop,timestamp 1295625 42799191>
(DF) [tos 0x10]
12:26:49.641808 mu7.matrix.sara.nl.20000 > teras.sara.nl.2811: F
5698:5698(0) ack 8558 win 20272 <nop,nop,timestamp 42799194 1295625>
(DF)
12:26:49.643452 teras.sara.nl.2811 > mu7.matrix.sara.nl.20000: . ack
5699 win 64512 <nop,nop,timestamp 1295625 42799194> (DF) [tos 0x10]
12:26:49.652814 mu7.matrix.sara.nl.20000 > teras.sara.nl.2811: S
2504656570:2504656570(0) win 5840 <mss 1460,sackOK,timestamp 42799195
0,nop,wscale 0> (DF)
12:26:49.653974 teras.sara.nl.2811 > mu7.matrix.sara.nl.20000: R
8558:8558(0) ack 5699 win 64512 (DF) [tos 0x10]
12:26:49.654017 teras.sara.nl.2811 > mu7.matrix.sara.nl.20000: R
3392287157:3392287157(0) ack 797379 win 0

The communication ends when the gridftp server (teras.sara.nl) sends two
RST (reset) packages. Any ideas?

Ron Trompert
<[log in to unmask]>