Обсуждение: URGENT - startup process (PID 29541) was terminated by signal 6
Our Opteron DB server had a problem with its RAID controller requiring an immediate shutdown of our Postgres server (8.1.3 on FreeBSD 6.0 release number 10). We used kill -QUIT on the postmaster PID. After repairing and rebooting the server we tried to start Postgres and get the following: 2006-10-06 12:32:40 CDT FATAL: the database system is starting up 2006-10-06 12:32:40 CDT FATAL: the database system is starting up 2006-10-06 12:32:40 CDT PANIC: heap_clean_redo: no block 2006-10-06 12:32:40 CDT LOG: startup process (PID 29541) was terminated by signal 6 2006-10-06 12:32:40 CDT LOG: aborting startup due to startup process failure At shutdown we had the following: 2006-10-06 12:08:48 CDT PANIC: right sibling is not next child in "winddir_idxu" 2006-10-06 12:08:48 CDT STATEMENT: VACUUM ANALYZE VERBOSE 2006-10-06 12:08:49 CDT LOG: server process (PID 91933) was terminated by signal 6 2006-10-06 12:08:49 CDT LOG: terminating any other active server processes Anything I can do? -------------------------------------------------------------------------------- Andrew Rost National Operational Hydrologic Remote Sensing Center (NOHRSC) National Weather Service, NOAA 1735 Lake Dr. West, Chanhassen, MN 55317-8582 Voice: (952)361-6610 x 234 Fax: (952)361-6634 andy.rost@noaa.gov http://www.nohrsc.noaa.gov --------------------------------------------------------------------------------
andy rost <Andy.Rost@noaa.gov> writes:
> Our Opteron DB server had a problem with its RAID controller requiring
> an immediate shutdown of our Postgres server (8.1.3 on FreeBSD 6.0
> release number 10). We used kill -QUIT on the postmaster PID.
> 2006-10-06 12:32:40 CDT PANIC: heap_clean_redo: no block
> ...
> 2006-10-06 12:08:48 CDT PANIC: right sibling is not next child in
> "winddir_idxu"
Um, were you running with full_page_writes off? Bad idea in 8.1 :-( ...
especially on hardware that turns out to not be 100% reliable.
I think your only hope of restarting the database is to use
pg_resetxlog. This is likely to leave you with some database
corruption, in the form of partially applied recent transactions.
I'd recommend a dump and reload, or at least REINDEXing all the
indexes, to forestall problems from that. While you are at it,
update to 8.1.4.
regards, tom lane
Hello Tom,
* Tom Lane <tgl@sss.pgh.pa.us>, [2006-10-06 15:16 -0400]:
> Um, were you running with full_page_writes off? Bad idea in 8.1 :-( ...
The manual [1] says that full_page_writes is ignored and always treated
as if it was set to on. Is it wrong?
ciao,
ema
[1] http://www.postgresql.org/docs/8.1/static/runtime-config-wal.html#GUC-FULL-PAGE-WRITES
Вложения
Emanuele Rocca wrote: -- Start of PGP signed section. > Hello Tom, > > * Tom Lane <tgl@sss.pgh.pa.us>, [2006-10-06 15:16 -0400]: > > Um, were you running with full_page_writes off? Bad idea in 8.1 :-( ... > > The manual [1] says that full_page_writes is ignored and always treated > as if it was set to on. Is it wrong? Well, early versions of 8.1.X did honor full_page_writes, but 8.1.4 disabled it. -- Bruce Momjian bruce@momjian.us EnterpriseDB http://www.enterprisedb.com + If your life is a hard drive, Christ can be your backup. +