Not so long ago, I explained here some problems encountered while FTP
mgetting a huge number of files on a SS1000 / Solaris 2.5 from a HPUX 10.20
client.
The symptom was erratic "425 Can't build data connection: Address already
in use." messages from the SS1000 during the FTP session and missing files
at the end on the HP.
Obviously, I already had place a call to our Sun Support.
The solution is that there's no solution with Solaris 2.5 nor in 2.5-1 and
2.6 (although the ftpd daemon changed).
Decreasing tcp_close_wait_interval to 30000 (30", was 240000, i.e. 4') had
no effect at all and implies all tcp connections.
I have been told to use wu-ftpd instead of the standard ftpd daemon
provided by Sun. But since the, I faced some reluctance here, because it is
not supposed to be officially supported...
Guess some will have to change their mind.
< Original message >
Hello there,
A HP server (under HPUX 10.20) accesses a SS1000/Sol 2.5 via FTP to perform
multiple gets (mget) over a very large number of files (+1000).
After a while, the in.ftpd daemon on the Sun sends :
<...>
200 PORT command successful.
425 Can't build data connection: Address already in use.
and the transfer hangs up for while (but sometimes it does not).
Other "425 Can't build data connection: Address already in use." lines may
be displayed along the mget session but it happens that it comes back to
normal operation.
I tried it with another Sun as FTP server : Ultra 30 / Solaris 2.6 instead
of the SS100/2.5. Same song !
What I could see on the Sun while a FTP get was hanging on is something like :
(netstat -a | grep ftp-data)
sun.ftp-data hp.3940 57344 0 24820 0 TIME_WAIT
sun.ftp-data hp.4333 57344 0 24820 0 TIME_WAIT
. . .
sun.ftp-data hp.1390 57344 0 24820 0 TIME_WAIT
sun.ftp-data hp.2682 0 0 24656 0 BOUND
Bound (they say) ?
I placed a call 2 weeks ago to the french Sun support and still get no
solution...
I also tried :
* HP client - HP server : hang up but no 425 blabla
* HP client - VMS/UCX 4.2 : hang up but no 425 blabla
* HP client - PC/W95/eXceed6 : no hang up, no 425 blabla
* Sun Client - Sun Server : no hang up, no 425 blabla
So, it looks like there's something strange between Solaris and HPUX.
A HP patch PHNE_13597 has been applied to the HP client but for no change.
Here's what shorev -p says on the SS1000 / Sol. 2.5 :
showrev: get_env_var(core, SUNW_PATCHID)
Sorry, but I did not find any FAQ/summary/comment/etc... about this
behaviour anywhere.
I've been looking around in the RFCs but found nothing very interesting.
Any clue ?
TIA
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:16 CDT