Обсуждение: error : missing chunk number 3 for toast value 14146136 in pg_toast_8895383
Hi All,
We are performing the PostgreSQL DB upgrade using pg_upgrade from 9.6 to 12.5 on the production server. The DB upgrade activity completed successfully on prod server for all databases.
After upgraded, we are performing logical backup & vacuum full on each DB and getting error only on 2 big tables and remaining DB working fine
ERROR: missing chunk number 3 for toast value 14146136 in pg_toast_8895383
ERROR: missing chunk number 29 for toast value 14674367 in pg_toast_8895400
As we go through the various blocks as well community old issues and performed the below steps however, still facing the same error.
REINDEX System mydb;
REINDEX table mytable;
REINDEX table pg_toast.pg_toast_8895383;
VACUUM analyze mytable;
or
VACUUM FULL ANALYZE mytable;
Also tried with drop these tables & recreate the same tables and load data from the backup file still getting the same error.
We verified there is no data corruption on 9.6 only facing issues above error after upgrade 12.5.
Please guide us to resolve this error.
Appreciate your help.
We are performing the PostgreSQL DB upgrade using pg_upgrade from 9.6 to 12.5 on the production server. The DB upgrade activity completed successfully on prod server for all databases.
After upgraded, we are performing logical backup & vacuum full on each DB and getting error only on 2 big tables and remaining DB working fine
ERROR: missing chunk number 3 for toast value 14146136 in pg_toast_8895383
ERROR: missing chunk number 29 for toast value 14674367 in pg_toast_8895400
As we go through the various blocks as well community old issues and performed the below steps however, still facing the same error.
REINDEX System mydb;
REINDEX table mytable;
REINDEX table pg_toast.pg_toast_8895383;
VACUUM analyze mytable;
or
VACUUM FULL ANALYZE mytable;
Also tried with drop these tables & recreate the same tables and load data from the backup file still getting the same error.
We verified there is no data corruption on 9.6 only facing issues above error after upgrade 12.5.
Please guide us to resolve this error.
Appreciate your help.
Thanks & Regards,
Manoj
Manoj
manoj kathar <manojkathar3@gmail.com> writes:
> We are performing the PostgreSQL DB upgrade using pg_upgrade from 9.6 to
> 12.5 on the production server.
What is the rationale for choosing 12.5? The current v12-branch release
is 12.11 (and 12.12 will be out next week). So you're missing nearly
two years' worth of bug fixes. A quick look through the release notes
shows at least one pg_upgrade fix that could be relevant, from 12.8:
Make <application>pg_upgrade</application> carry forward the old
installation's <literal>oldestXID</literal> value (Bertrand Drouvot)
regards, tom lane
Re: error : missing chunk number 3 for toast value 14146136 in pg_toast_8895383
От
manoj kathar
Дата:
Thanks for your response.
We installed the PostgreSQL-12.11 version and performed DB upgrade activity, still facing the same error on the same 2 tables.
We performed the below steps to resolve the error, however, no luck.
1. Upgrade All DB's to 12.11 from 9.6 and successfully completed.
2. Performed the vacuum full analyze on upgrade data i.e 12.11, it's also completed successfully.
3. Performed the Logical dump backup on all DB's, failed only 2 tables due to chunk error.
4. Performed testing with dropping these 2 tables & recreating and loading data from backup file (copy command), loading data completed without any issues, however, facing issues for data selection as well as logical backup.
5. Performed testing with truncating table , loading data from backup file, loading data completed without any issues, however, facing issues for data selection.
6. Deleted only corrupted data & performed the same activities again getting the same error.
These 2 tables are big in size and having bytea, text columns might be this column resulting in a chunk error.
Is there any other way to resolve this.
We installed the PostgreSQL-12.11 version and performed DB upgrade activity, still facing the same error on the same 2 tables.
We performed the below steps to resolve the error, however, no luck.
1. Upgrade All DB's to 12.11 from 9.6 and successfully completed.
2. Performed the vacuum full analyze on upgrade data i.e 12.11, it's also completed successfully.
3. Performed the Logical dump backup on all DB's, failed only 2 tables due to chunk error.
4. Performed testing with dropping these 2 tables & recreating and loading data from backup file (copy command), loading data completed without any issues, however, facing issues for data selection as well as logical backup.
5. Performed testing with truncating table , loading data from backup file, loading data completed without any issues, however, facing issues for data selection.
6. Deleted only corrupted data & performed the same activities again getting the same error.
These 2 tables are big in size and having bytea, text columns might be this column resulting in a chunk error.
Is there any other way to resolve this.
Thanks & Regards,
Manoj
Manoj
On Sat, Aug 6, 2022 at 10:38 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
manoj kathar <manojkathar3@gmail.com> writes:
> We are performing the PostgreSQL DB upgrade using pg_upgrade from 9.6 to
> 12.5 on the production server.
What is the rationale for choosing 12.5? The current v12-branch release
is 12.11 (and 12.12 will be out next week). So you're missing nearly
two years' worth of bug fixes. A quick look through the release notes
shows at least one pg_upgrade fix that could be relevant, from 12.8:
Make <application>pg_upgrade</application> carry forward the old
installation's <literal>oldestXID</literal> value (Bertrand Drouvot)
regards, tom lane