Обсуждение: Postgresql12: ERROR: Could not read from file "pg_act/02 F 4" at offset 253952: read too few bytes
Postgresql12: ERROR: Could not read from file "pg_act/02 F 4" at offset 253952: read too few bytes
Hello.
While the postgres database was running, we encountered the following error:
ERROR: could not access status of transaction 793754117 Detail: Could not read from file "pg_act/02 F 4" at offset 253952: read too few bytes.
The pg_act/02 F 4 file on the main database had a different size from other files. At the same time, the pg_xact/02F4 file on the backup database had a standard size of 256Kb. There are no errors writing to the file system. No errors were found in the postgres logs either.
After this error occurred, working with the main database was impossible, and we had to switch to the backup database.
What could have caused this error in postgres? Was it possible to restore the database without switching to a backup? How can I prevent such errors from occurring in the future? How to correctly troubleshooting the cause of this error?
Postgres version: 12.1-2
OS: Oracle Linux Server release 7.6
With Regards,
Aleksey.
RE: Postgresql12: ERROR: Could not read from file "pg_act/02 F 4" at offset 253952: read too few bytes
Hello.
Do you have a result on my problem? This is critical for us.
From: Orishich Aleksey <a.orishich@cft.ru>
Sent: Wednesday, July 14, 2021 10:03 AM
To: pgsql-bugs@lists.postgresql.org
Subject: [MASSMAIL] Postgresql12: ERROR: Could not read from file "pg_act/02 F 4" at offset 253952: read too few bytes
Hello.
While the postgres database was running, we encountered the following error:
ERROR: could not access status of transaction 793754117 Detail: Could not read from file "pg_act/02 F 4" at offset 253952: read too few bytes.
The pg_act/02 F 4 file on the main database had a different size from other files. At the same time, the pg_xact/02F4 file on the backup database had a standard size of 256Kb. There are no errors writing to the file system. No errors were found in the postgres logs either.
After this error occurred, working with the main database was impossible, and we had to switch to the backup database.
What could have caused this error in postgres? Was it possible to restore the database without switching to a backup? How can I prevent such errors from occurring in the future? How to correctly troubleshooting the cause of this error?
Postgres version: 12.1-2
OS: Oracle Linux Server release 7.6
With Regards,
Aleksey.
Re: Postgresql12: ERROR: Could not read from file "pg_act/02 F 4" at offset 253952: read too few bytes
> On 27 Jul 2021, at 13:38, Orishich Aleksey <a.orishich@cft.ru> wrote: > Do you have a result on my problem? Troubleshooting why a file on your filesystem was corrupt/torn after the fact when no logs or traces of evidence exists is next to impossible. In this case there isn't even an errno set from the system. A good place to start would be to upgrade your system as you are several bugfix releases behind. > This is critical for us. The PostgreSQL website lists several professional services companies who offer support contracts with SLA's which this list doesn't, and who can be hands-on to investigate should this ever happen again. -- Daniel Gustafsson https://vmware.com/
RE: Postgresql12: ERROR: Could not read from file "pg_act/02 F 4" at offset 253952: read too few bytes
ОК, what logs do you need? -----Original Message----- From: Daniel Gustafsson <daniel@yesql.se> Sent: Tuesday, July 27, 2021 6:58 PM To: Оришич Алексей Петрович <a.orishich@cft.ru> Cc: pgsql-bugs@lists.postgresql.org Subject: Re: Postgresql12: ERROR: Could not read from file "pg_act/02 F 4" at offset 253952: read too few bytes > On 27 Jul 2021, at 13:38, Orishich Aleksey <a.orishich@cft.ru> wrote: > Do you have a result on my problem? Troubleshooting why a file on your filesystem was corrupt/torn after the fact when no logs or traces of evidence exists isnext to impossible. In this case there isn't even an errno set from the system. A good place to start would be to upgradeyour system as you are several bugfix releases behind. > This is critical for us. The PostgreSQL website lists several professional services companies who offer support contracts with SLA's which this listdoesn't, and who can be hands-on to investigate should this ever happen again. -- Daniel Gustafsson https://vmware.com/