Re: Weaker shmem interlock w/o postmaster.pid
Robert Haas <robertmhaas@gmail.com>
On Tue, Sep 10, 2013 at 11:33 PM, Noah Misch <noah@leadboat.com> wrote: > I'm thinking to preserve postmaster.pid at immediate shutdown in all released > versions, but I'm less sure about back-patching a change to make > PGSharedMemoryCreate() pickier. On the one hand, allowing startup to proceed > with backends still active in the same data directory is a corruption hazard. > On the other hand, it could break weird shutdown/restart patterns that permit > trivial lifespan overlap between backends of different postmasters. Opinions? I'm more sanguine about the second change than the first. Leaving postmaster.pid around seems like a clear user-visible behavior change that could break user scripts or have other consequences that we don't foresee; thus, I would vote against back-patching it. Indeed, I'm not sure it's a good idea to do that even in master. On the other hand, tightening the checks in PGSharedMemoryCreate() seems very much worth doing, and I think it might also be safe enough to back-patch. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company
Commits
-
Probe only 127.0.0.1 when looking for ports on Unix.
- 1ed42e50e963 9.6.14 landed
- 8c7a8e19bb17 10.9 landed
- 1f3bcb497200 11.4 landed
- 8efe710d9c84 12.0 landed
-
Don't write to stdin of a test process that could have already exited.
- e12a4726127b 12.0 cited
-
Test both 0.0.0.0 and 127.0.0.x addresses to find a usable port.
- 9f65081cf74b 9.6.13 landed
- 4543ef36f0ef 10.8 landed
- 0bdf6d635e42 11.3 landed
- 4ab02e815627 12.0 landed
-
MSYS: Translate REGRESS_SHLIB to a Windows file name.
- 31e2caaceea9 11.3 landed
- 9daefff12260 12.0 landed
-
MSYS: Skip src/test/recovery/t/017_shm.pl.
- 6b7906e1d9e7 9.6.13 landed
- 2bc0474792e3 10.8 landed
-
When Perl "kill(9, ...)" fails, try "pg_ctl kill".
- de262941fc9c 11.3 landed
- 61c0962d90d5 10.8 landed
- 55926b0a0917 9.6.13 landed
- 947a35014fdc 12.0 landed
-
Consistently test for in-use shared memory.
- 3ef5e16c6181 9.4.22 landed
- c565de643167 9.5.17 landed
- 808e1e75fb37 9.6.13 landed
- 7ef2b313e649 11.3 landed
- 6d81e3c6526b 10.8 landed
- c098509927f9 12.0 landed
- e7f89a5dec0f 9.5.17 landed
- b2307f8e3184 11.3 landed
- 891e9e0aee53 9.6.13 landed
- 7c414cdc39a3 10.8 landed
- 3a70b66e6257 9.4.22 landed
- 2f932f71d9f2 12.0 landed
-
Revert "Consistently test for in-use shared memory."
- ae98c24fda4b 9.6.13 landed
- 9ec582b64d69 9.5.17 landed
- 7d18a55c90eb 10.8 landed
- 392ea22e9b32 11.3 landed
- 0777696fe346 9.4.22 landed
- 82150a05be0d 12.0 landed
-
Silence -Wimplicit-fallthrough in sysv_shmem.c.
- 0aa0ccfdd653 9.6.13 cited
-
Make src/test/recovery/t/017_shm.pl safe for concurrent execution.
- 0e0d449fb02a 9.6.13 landed
- 3186d5fac55b 10.8 landed
- 1106438c37db 11.3 landed
- 16ee6eaf80a4 12.0 landed
-
Update HINT for pre-existing shared memory block.
- f3461cb8f5b5 9.4.22 landed
- 52e7e4d1dfe6 9.5.17 landed
- e49dc9ab14b2 9.6.13 landed
- ab7590e9197c 11.3 landed
- 327399412d03 10.8 landed
- 5a907404b527 12.0 landed
-
Add WL_EXIT_ON_PM_DEATH pseudo-event.
- cfdf4dc4fc96 12.0 cited
-
The default values for shared_buffers and max_connections are now 1000
- de98a7e23a19 7.4.1 cited
-
XLOG (and related) changes:
- 4d14fe0048cf 7.1.1 cited
-
Significant cleanups in SysV IPC handling (shared mem and semaphores).
- c715fdea2678 7.1.1 cited