Putting a 256kb file full of 0x55 that's 01010101 and represents 4 commits It did the job in being able to restart the server. According to our data a “better” way, with less garbage.
The “Toast” issues how ever are still present.
To spend our weekend well we setup a new server with version 12.1 but had to fallback on 11.6 ( see other post )
We kept our “old” server active to see if we can learn some more from this hard-times.
Thanks for the help
Marc
On 5 Feb 2020, at 12:14, Nick Renders wrote:
Hello,
Yesterday, we experienced some issues with our Postgres installation (v9.6 running on macOS 10.12). It seems that the machine was automatically rebooted for a yet unknown reason, and afterwards we were unable to start the Postgres service.
The postgres log shows the following:
2020-02-04 15:20:41 CET LOG: database system was interrupted; last known up at 2020-02-04 15:18:34 CET 2020-02-04 15:20:43 CET LOG: database system was not properly shut down; automatic recovery in progress 2020-02-04 15:20:44 CET LOG: invalid record length at 14A/9E426DF8: wanted 24, got 0 2020-02-04 15:20:44 CET LOG: redo is not required 2020-02-04 15:20:44 CET FATAL: could not access status of transaction 247890764 2020-02-04 15:20:44 CET DETAIL: Could not read from file "pg_clog/00EC" at offset 106496: Undefined error: 0. 2020-02-04 15:20:44 CET LOG: startup process (PID 403) exited with exit code 1 2020-02-04 15:20:44 CET LOG: aborting startup due to startup process failure 2020-02-04 15:20:44 CET LOG: database system is shut down
After some searching, I found someone who had had a similar issue and was able to resolve it by overwriting the file in pg_clog. So I tried the following command:
But I am worried that there might still be some issues that we haven't noticed yet. I also have no idea what caused this error in the first place. It might have been the reboot, but maybe the reboot was a result of a Postgres issue.
Is there anything specific I should check in our postgres installation / database to make sure it is running ok now? Anyway to see what the consequences were of purging that one pg_clog file?
