Re: BUG #8579: CoreDump of background writer process

alvherre(at)2ndquadrant(dot)com wrote:

> I thought we no longer supported QNX ... in fact, I think we removed it before the 8.2 release. So what's this port you're using? This probably means a heavily patched source.

Yes we had to add some patches, but not as much as you are expecting. We were running the 8.3 release with the same patches for a long time without these kinds of troubles. I don't had to add much more patches to get the new release running and passing the regression-test successfully .

Patched Files:

patch-configure
patch-configure.in
patch-contrib_Makefile
patch-src_Makefile.shlib
patch-src_backend_Makefile
patch-src_backend_access_common_reloptions.c
patch-src_backend_libpq_pqsignal.c
patch-src_backend_port_Makefile
patch-src_backend_port_ipc__test.c
patch-src_backend_port_sysv__shmem.c
patch-src_backend_postmaster_postmaster.c
patch-src_backend_replication_libpqwalreceiver_Makefile
patch-src_backend_snowball_Makefile
patch-src_backend_tcop_postgres.c
patch-src_backend_utils_init_postinit.c
patch-src_backend_utils_mb_conversion__procs_proc.mk
patch-src_backend_utils_misc_guc.c
patch-src_backend_utils_misc_ps__status.c
patch-src_bin_pg__dump_pg__backup.h
patch-src_bin_pg__dump_pg__backup__archiver.c
patch-src_bin_pg__dump_pg__backup__archiver.h
patch-src_bin_pg__dump_pg__backup__tar.c
patch-src_bin_pg__dump_pg__dump.c
patch-src_bin_pg__dump_pg__dump.h
patch-src_bin_pg__dump_pg__dumpall.c
patch-src_include_pg__config__manual.h
patch-src_include_regex_regguts.h
patch-src_include_storage_s__lock.h
patch-src_include_tcop_tcopprot.h
patch-src_interfaces_ecpg_ecpglib_misc.c
patch-src_interfaces_ecpg_include_ecpglib.h
patch-src_interfaces_libpq_pqsignal.c
patch-src_pl_plpgsql_src_Makefile
patch-src_port_isinf.c
patch-src_test_regress_pg__regress.c
patch-src_test_regress_resultmap
patch-src_timezone_localtime.c
patch-src_timezone_private.h
patch-src_timezone_strftime.c
patch-src_timezone_zic.c

Files to be added:

GNUmakefile.libpq
contrib_dbmirrorc_AddTrigger.sql
contrib_dbmirrorc_DBMirror.pl
contrib_dbmirrorc_Makefile
contrib_dbmirrorc_MirrorSetup.sql
contrib_dbmirrorc_README.dbmirror
contrib_dbmirrorc_clean__pending.pl
contrib_dbmirrorc_pending.c
contrib_dbmirrorc_slaveDatabase.conf
dynloader-ltdl.h
src_backend_port_dynloader_qnx6.c
src_backend_port_dynloader_qnx6.h
src_backend_port_qnx6_Makefile
src_backend_port_qnx6_shm.c
src_backend_port_qnx6_shm.h
src_include_port_qnx6.h
src_makefiles_Makefile.qnx6
src_template_qnx6

> This is also a bit funny. Normally each segment is 1GB long, unless you configure it otherwise; this means a segment has only 131072 blocks.
> But you have at least 149256 blocks here.

These are the changes from postgresql.conf, all other settings are commented-out:

unix_socket_directory = '/bac/sys/data'
tcp_keepalives_idle = 120
max_files_per_process = 200
vacuum_cost_delay = 20ms
checkpoint_segments = 6
log_destination = 'stderr'
logging_collector = on
log_directory = '/bac/sys/log'
log_filename = 'postgresql.log.%a'
log_truncate_on_rotation = on
log_rotation_age = 1d
log_rotation_size = 0
log_min_duration_statement = 60000
log_hostname = on
log_line_prefix = '%a-%d-%h-%p-%t:'
autovacuum = on
search_path = 'public,syscore,systab000,gltcore,glttab001'
bytea_output = 'escape'
datestyle = 'iso, mdy'
lc_messages = 'POSIX'
lc_monetary = 'POSIX'
lc_numeric = 'POSIX'
lc_time = 'POSIX'
max_locks_per_transaction = 100

I hope this helps...

Mit freundlichen Grüßen aus Krefeld,
With best regards from Krefeld,

CS-Lab GmbH
i. A. René Grün

E-Mail: rgr(at)cslab(dot)de
Fon:    +49 2151 72949-0
Fax:    +49 2151 72949-9
---
CS-Lab GmbH (Creativ Software Labor GmbH)
Römerstr. 15
D-47809 Krefeld
Geschäftsführer: Dieter Schmitz
Registergericht Krefeld, HRB 12257, USt.-ID: DE 263 834 180