I've been testing pg_standby as a helper application for a warm standby setup. So far, so good. When the environment is controlled and everything happens as expected, I'm able to operate a basic primary/standby setup. (This is all using 8.2.3 on Solaris x86, btw.) One thing I noticed in early testing, though, was the scenario where the trigger file can't be removed by pg_standby. I touched a trigger file as root, which made it unremovable by postgres. So this tripped the relevant error condition in pg_standby. I had a little difficulty understanding in what state this left the recovery process, and I'm not helping myself much by reading the code. Doesn't the non-zero exit from CheckForExternalTrigger mean that pg_standby will be signaling to the standby server a file-not-found scenario? -- Thomas F. O'Connell optimizing modern web applications : for search engines, for usability, and for performance : |
