Thread: error 57014
Hello.
The next query runs into error 57014 in about 50 % of the cases. We certainly have not aborted the request. In the other 50 % it runs OK. The statement is executed
in pgsql 7.4.8
Statement:
------------
SELECT pst.typ_stay, SUM(pst.nr_rooms), SUM(pst.nr_trans), SUM(pst.pst_amnt) from pst where pst.companyc = 'H' AND pst.pst_date >= (date '2006-01-01') AND pst.pst_date <= (date '2006-06-08') AND pst.pst_type IN (1,7) AND pst.typ_stay IS NOT NULL AND pst.dpt_code = 'FO' GROUP BY pst.typ_stay ;
Extract logfile
----------------
2006-06-09 03:55:06 LOG: 00000: database system is ready
LOCATION: StartupXLOG, xlog.c:2946
2006-06-09 03:55:37 ERROR: 57014: canceling query due to user request
LOCATION: ProcessInterrupts, postgres.c:1964
STATEMENT: SELECT pst.typ_stay, SUM(pst.nr_rooms), SUM(pst.nr_trans), SUM(pst.pst_amnt) from pst where pst.companyc = 'H' AND pst.pst_date >= (date '2006-01-01') AND pst.pst_date <= (date '2006-06-08') AND pst.pst_type IN (1,7) AND pst.typ_stay IS NOT NULL AND pst.dpt_code = 'FO' GROUP BY pst.typ_stay ;
2006-06-09 03:56:07 LOG: 08P01: unexpected EOF on client connection
LOCATION: StartupXLOG, xlog.c:2946
2006-06-09 03:55:37 ERROR: 57014: canceling query due to user request
LOCATION: ProcessInterrupts, postgres.c:1964
STATEMENT: SELECT pst.typ_stay, SUM(pst.nr_rooms), SUM(pst.nr_trans), SUM(pst.pst_amnt) from pst where pst.companyc = 'H' AND pst.pst_date >= (date '2006-01-01') AND pst.pst_date <= (date '2006-06-08') AND pst.pst_type IN (1,7) AND pst.typ_stay IS NOT NULL AND pst.dpt_code = 'FO' GROUP BY pst.typ_stay ;
2006-06-09 03:56:07 LOG: 08P01: unexpected EOF on client connection
Anyone any idea?
Many thanks
Henk Sanders
"H.J. Sanders" <hjs@rmax.nl> writes: > The next query runs into error 57014 in about 50 % of the cases. We > certainly have not aborted the request. In the other 50 % it runs OK. The > statement is executed > in pgsql 7.4.8 Well, *something* sent that backend a SIGINT. You should look around for possible causes. I've heard that some kernels use SIGINT for resource-limit kills (ie, ulimit exceeded). regards, tom lane
"Tom Lane" <tgl@sss.pgh.pa.us> wrote > > Well, *something* sent that backend a SIGINT. > Could something be the backend statement_timeout? I remember [but not sure] we don't differenciate timeout or explicite SIGINT in 7.4.x. Regards, Qingqing