Обсуждение: interesting SIGNAL 10 (BUSERR) on CVS
running 7.4RC2+ (from last week). I got the following: $ debug -ic -c core* /usr/local/pgsql/bin/postgres Warning: No debugging information in /usr/local/pgsql/bin/postgres Core image of postgres (process p1) created CORE FILE [_bt_getroot] SIGNALED 10 (bus code[BUS_OBJERR] address[0xbb12c550]) in p1 0x808f057 (_bt_getroot+135:) movzwl 16(%edi),%edx debug> trace ^ Error: Unrecognized keyword at trace debug> stack Stack Trace for p1, Program postgres *[0] _bt_getroot(0x83520dc, 0x1) [0x808f057][1] _bt_search(presumed: 0x83520dc, 0x1, 0x838560c) [0x809350d][2]_bt_first(presumed: 0x83853d8, 0x1, 0x8385400) [0x809443e][3] btgettuple() [0x8091fa6] debug> It's a one shot, and I have a core, but, I don't have the debug symbols as UnixWare's cc doesn't allow -g and -O. Do we care? -- Larry Rosenman http://www.lerctr.org/~ler Phone: +1 972-414-9812 E-Mail: ler@lerctr.org US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749
Larry Rosenman <ler@lerctr.org> writes:
> I got the following:
> $ debug -ic -c core* /usr/local/pgsql/bin/postgres
> Warning: No debugging information in /usr/local/pgsql/bin/postgres
> Core image of postgres (process p1) created
> CORE FILE [_bt_getroot]
> SIGNALED 10 (bus code[BUS_OBJERR] address[0xbb12c550]) in p1
> 0x808f057 (_bt_getroot+135:) movzwl 16(%edi),%edx
Is it repeatable? It's hard to see how _bt_getroot() could core
except maybe in the presence of serious data corruption in the index ...
regards, tom lane
--On Sunday, November 16, 2003 12:59:43 -0500 Tom Lane <tgl@sss.pgh.pa.us> wrote: > Larry Rosenman <ler@lerctr.org> writes: >> I got the following: > >> $ debug -ic -c core* /usr/local/pgsql/bin/postgres >> Warning: No debugging information in /usr/local/pgsql/bin/postgres >> Core image of postgres (process p1) created >> CORE FILE [_bt_getroot] >> SIGNALED 10 (bus code[BUS_OBJERR] address[0xbb12c550]) in p1 >> 0x808f057 (_bt_getroot+135:) movzwl 16(%edi),%edx > > Is it repeatable? It's hard to see how _bt_getroot() could core > except maybe in the presence of serious data corruption in the index ... it happened once, and postgres did a restart. The cronjob that triggered it (WebCalendar's sendreminders script), has run since. So, I don't know. I just thought it was weird. LER -- Larry Rosenman http://www.lerctr.org/~ler Phone: +1 972-414-9812 E-Mail: ler@lerctr.org US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749
Larry Rosenman <ler@lerctr.org> writes:
>> Is it repeatable? It's hard to see how _bt_getroot() could core
>> except maybe in the presence of serious data corruption in the index ...
> it happened once, and postgres did a restart. The cronjob that triggered
> it (WebCalendar's sendreminders script), has run since. So, I don't know.
Hmm. It might be worth running memtest86 to look for flaky RAM.
I doubt it's worth the effort to try to extract any info from a core
file with no symbols :-(
regards, tom lane
--On Sunday, November 16, 2003 13:15:27 -0500 Tom Lane <tgl@sss.pgh.pa.us> wrote: > Larry Rosenman <ler@lerctr.org> writes: >>> Is it repeatable? It's hard to see how _bt_getroot() could core >>> except maybe in the presence of serious data corruption in the index ... > >> it happened once, and postgres did a restart. The cronjob that triggered >> it (WebCalendar's sendreminders script), has run since. So, I don't >> know. > > Hmm. It might be worth running memtest86 to look for flaky RAM. > I doubt it's worth the effort to try to extract any info from a core > file with no symbols :-( Okie. I just figured I'd give a heads up that it happened. -- Larry Rosenman http://www.lerctr.org/~ler Phone: +1 972-414-9812 E-Mail: ler@lerctr.org US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749