Discussion:
sporadic BCP aborts with message: Interrupt - Bulk copy aborted.
(too old to reply)
b***@msn.com
2011-09-29 22:52:20 UTC
Permalink
Hello,
We run a lot of Linux and SunOS Sybase installations of version 12.5.4 and
15.0.3 ESD#4 and encounter in some SunOS installtions the following sporadic
- table is truncated and all Indices are dropped
  bcp $DATABASE.owner.$tabelle in $file -t '^A' -b 10000 -Uxxxxx -Pxxxxxxxxxx -c --maxconn 1
...
9290000 Zeilen an den Adaptive Server gesandt.
Import der Daten in den Adaptive Server gem der Anweisungsfolge erfolgreich durchgef hrt.
9300000 Zeilen an den Adaptive Server gesandt.
Unterbrechung - Bulk-Copy abgebrochen.
(sorry for the German messages; it is sending over a lot of lines
successfull, and than terminates in the "midlle" of the file with
Interrupt - Bulk copy aborted.)
it does not depend on the data, because you can truncate the table again
and fire up the same BCP command, sometimes only successfull in the 3rd
attempt; most of the time it works fine in the 1st run;
as I said, this happens only on SunOS, not on SLES Linux; the ASE server
is local on the same host;
Could someone with access to the source code, please check what reasons
trigger such an abort? Any other idea?
Thanks in advance
        matthias
The usual reason for the message "Interrupt - Bulk copy aborted"
is the person running BCP doing a ^c to interrupt the process.
Perhaps something else on the system is sending a SIGINT signal
(randomly?) to other processes. You might be able to catch this
happening/protect against it using the "trap" unix command.
Example: http://steve-parker.org/sh/trap.shtml
rebelde
2011-09-30 08:19:32 UTC
Permalink
Post by b***@msn.com
The usual reason for the message "Interrupt - Bulk copy aborted"
is the person running BCP doing a ^c to interrupt the process.
Perhaps something else on the system is sending a SIGINT signal
(randomly?) to other processes.  You might be able to catch this
happening/protect against it using the "trap" unix command.
Example: http://steve-parker.org/sh/trap.shtml
Hello Bret,

Thanks for your hints. I think, Ctrl-C is not the reason, because the BCP is
started by some script which in turn is started with 'nohup'. Ofc, a
SIGINT fired up by some other proc / session is possible.

I will try to start the BCP under truss(1) to catch the reason.

matthias
Derek Asirvadem
2011-10-15 14:07:59 UTC
Permalink
Matthias

Reduce the batch size. 10,000 is too large. See if the error
continues to occur.

Regards
Derek

Loading...