[Info-vax] SCP2 and large files

[Jeremy Begg]

Hi!

A customer running HP OpenVMS V8.4 with TCP/IP Services V5.7-ECO5 is 
having a problem with SCP2.

The nightly VMS backup job creates BACKUP savesets on a local disk.
When each saveset is completed (i.e. when the BACKUP/IMAGE finishes) a 
batch job is submitted to transfer the saveset to a Linux system using 
SCP2.  (From there it goes to a large archiving system.)

Those transfers work fine for most of the savesets, which vary in size 
from 9GB to 52GB.  However there's one saveset which is close to 500GB 
for which the transfer never works.

It appears that SCP2 does the following:

1. Connects to the remote (Linux) server and logs in.
2. Does a stat() on the saveset (presumably to get its exact size).
3. Starts transferring the file.

I think where it's going wrong is step 2.  The stat() on the 500GB 
saveset takes a very long time (seems to be about 90 minutes), by which 
time the Linux end has given up waiting for the transfer to begin.

Therefore two possible solutions spring to mind:

a. Can the CRTL stat() routine be told to guess the file size from the 
RMS file header (ALQ perhaps)?  I thought there might be a DECC$ feature 
logical to do this, but I can't find it.

b. Can SCP2 be told to enable TCP keepalive on the connection?

Thanks,

	Jeremy Begg