Обсуждение: pgsql: Get rid of the dynamic shared memory state file.
Get rid of the dynamic shared memory state file. Instead of storing the ID of the dynamic shared memory control segment in a file within the data directory, store it in the main control segment. This avoids a number of nasty corner cases, most seriously that doing an online backup and then using it on the same machine (e.g. to fire up a standby) would result in the standby clobbering all of the master's dynamic shared memory segments. Per complaints from Heikki Linnakangas, Fujii Masao, and Tom Lane. Branch ------ master Details ------- http://git.postgresql.org/pg/commitdiff/11a65eed1637a05b03e174700799b024e104bfb4 Modified Files -------------- src/backend/port/sysv_shmem.c | 12 ++- src/backend/port/win32_shmem.c | 6 +- src/backend/storage/ipc/dsm.c | 200 ++++++++-------------------------------- src/backend/storage/ipc/ipci.c | 6 +- src/include/storage/dsm.h | 8 +- src/include/storage/pg_shmem.h | 5 +- 6 files changed, 71 insertions(+), 166 deletions(-)
On 2014-04-08 15:45:39 +0000, Robert Haas wrote: > Get rid of the dynamic shared memory state file. > > Instead of storing the ID of the dynamic shared memory control > segment in a file within the data directory, store it in the main > control segment. This avoids a number of nasty corner cases, > most seriously that doing an online backup and then using it on > the same machine (e.g. to fire up a standby) would result in the > standby clobbering all of the master's dynamic shared memory > segments. > > Per complaints from Heikki Linnakangas, Fujii Masao, and Tom > Lane. Looks like it's not working for windows yet: http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=bowerbird&dt=2014-04-08%2016%3A31%3A18 http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=baiji&dt=2014-04-08%2017%3A00%3A01 There's also the to_reg* commit in the ones tested by those runs, but it sounds unlikely that those would trigger during initdb. Greetings, Andres Freund -- Andres Freund http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
On Tue, Apr 8, 2014 at 4:14 PM, Andres Freund <andres@2ndquadrant.com> wrote: > On 2014-04-08 15:45:39 +0000, Robert Haas wrote: >> Get rid of the dynamic shared memory state file. >> >> Instead of storing the ID of the dynamic shared memory control >> segment in a file within the data directory, store it in the main >> control segment. This avoids a number of nasty corner cases, >> most seriously that doing an online backup and then using it on >> the same machine (e.g. to fire up a standby) would result in the >> standby clobbering all of the master's dynamic shared memory >> segments. >> >> Per complaints from Heikki Linnakangas, Fujii Masao, and Tom >> Lane. > > Looks like it's not working for windows yet: > http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=bowerbird&dt=2014-04-08%2016%3A31%3A18 > http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=baiji&dt=2014-04-08%2017%3A00%3A01 > > There's also the to_reg* commit in the ones tested by those runs, but it > sounds unlikely that those would trigger during initdb. Crap. I did test that it worked with EXEC_BACKEND, but I don't have a Windows environment set up. I can try to eyeball the code and see what might be wrong with it, but... -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company
Robert Haas <rhaas@postgresql.org> writes:
> Get rid of the dynamic shared memory state file.
The Windows buildfarm critters don't seem to like this patch.
narwhal for example:
creating directory C:/msys/1.0/local/pgbuildfarm/buildroot/HEAD/pgsql.3944/src/test/regress/./tmp_check/data ... ok
creating subdirectories ... ok
selecting default max_connections ... 100
selecting default shared_buffers ... 128MB
selecting dynamic shared memory implementation ... windows
creating configuration files ... ok
creating template1 database in
C:/msys/1.0/local/pgbuildfarm/buildroot/HEAD/pgsql.3944/src/test/regress/./tmp_check/data/base/1... LOG: dynamic
sharedmemory control segment is corrupt
child process was terminated by exception 0xC0000005
baiji and bowerbird also crashed in initdb, though without the message
about "dynamic shared memory control segment is corrupt". Presumably
it's the same bug though.
regards, tom lane
On 2014-04-08 16:17:46 -0400, Robert Haas wrote: > On Tue, Apr 8, 2014 at 4:14 PM, Andres Freund <andres@2ndquadrant.com> wrote: > > On 2014-04-08 15:45:39 +0000, Robert Haas wrote: > >> Get rid of the dynamic shared memory state file. > >> > >> Instead of storing the ID of the dynamic shared memory control > >> segment in a file within the data directory, store it in the main > >> control segment. This avoids a number of nasty corner cases, > >> most seriously that doing an online backup and then using it on > >> the same machine (e.g. to fire up a standby) would result in the > >> standby clobbering all of the master's dynamic shared memory > >> segments. > >> > >> Per complaints from Heikki Linnakangas, Fujii Masao, and Tom > >> Lane. > > > > Looks like it's not working for windows yet: > > http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=bowerbird&dt=2014-04-08%2016%3A31%3A18 > > http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=baiji&dt=2014-04-08%2017%3A00%3A01 > > > > There's also the to_reg* commit in the ones tested by those runs, but it > > sounds unlikely that those would trigger during initdb. > > Crap. I did test that it worked with EXEC_BACKEND, but I don't have a > Windows environment set up. I can try to eyeball the code and see > what might be wrong with it, but... Not a windows guy either, but I just noticed that http://buildfarm.postgresql.org/cgi-bin/show_stage_log.pl?nm=narwhal&dt=2014-04-08%2016%3A00%3A03&stg=check has an additional log entry, the others don't: LOG: dynamic shared memory control segment is corrupt since that's apparently the first crash on narwahl related to this, it might be a hint where things are going wrong. Greetings, Andres Freund -- Andres Freund http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services