Thread: BUG #1443: Can't start service of Postgres
The following bug has been logged online: Bug reference: 1443 Logged by: phucle Email address: le.viet.phuc@gmail.com PostgreSQL version: 8.0 Operating system: Windows XP Pro japanese Description: Can't start service of Postgres Details: Dear, I installed Postgres 8.0 sucessfully and ran Ok but when I restarted my computer, Postgres could not start. When I checked in windows service, I realized that Postgres service did not start. I tried to start manual but could not start I have tested this problem many times. phucle
>The following bug has been logged online: > >Bug reference: 1443 >Logged by: phucle >Email address: le.viet.phuc@gmail.com >PostgreSQL version: 8.0 >Operating system: Windows XP Pro japanese >Description: Can't start service of Postgres >Details:=20 > >Dear, >=20 > I installed Postgres 8.0 sucessfully and ran Ok but when I=20 >restarted my >computer, Postgres could not start. When I checked in windows=20 >service, I >realized that Postgres service did not start. I tried to start=20 >manual but >could not start > I have tested this problem many times. You need to look in your server logs (in the pg_log directory) for hints about what's wrong. //Magnus
"phucle" wrote: > I installed Postgres 8.0 sucessfully and ran Ok but when I restarted my >computer, Postgres could not start. When I checked in windows service, I >realized that Postgres service did not start. I tried to start manual but >could not start I encoutered a similar problem on an laptop computer with an defective network adapter plug. Sometimes the defective plug leads to lost of the software network interface and this forces postgres to hang. OK, that's not a problem of postgre, but this accident unhides a postgre-bug. The hanging postgre instance isn't able to delete the it's lockfile postmaster.pid when shutting down. The further existence of this file will avoid any start of the pgsql service in the future, even after a reboot. You've to delete this file manually. I think that this is a bug. Not the simple existence of a lockfile should prohibit instances from run, but only the existence of an already opened lockfile.