Обсуждение: ERROR could not read block xxx in file, read only 0 of 8192 bytes
We have a few Postgres 16.2 servers running on VMs (RHEL 9.3) created on VMware vSphere (7.0.3). 3PAR LUNs are used as their storages.
On a couple of the PostgreSQL servers, there are errors such as:
ERROR: could not read block 1264 in file "base/16520/59258": read only 0 of 8192 bytes
The errors occured when trying to insert rows to the tables.
The blocks belong to tables. For instance,
select relname, relkind from pg_class where relfilenode = '59258';
relname | relkind
---------------+---------
smallgroupset | r`
I can pg_dump the all the rows of the table smallgroupset and there is no data loss.
pg_dump -d tabula -t smallgroupset -f smallgroupset.sql -v > smallgroupset.log 2>&1
I've tried to reindex the table ( and the whole database) but it made no difference.
Can anyone advise the root cause of the errors?
Any help will be much appreciated.
orclp
On a couple of the PostgreSQL servers, there are errors such as:
ERROR: could not read block 1264 in file "base/16520/59258": read only 0 of 8192 bytes
The errors occured when trying to insert rows to the tables.
The blocks belong to tables. For instance,
select relname, relkind from pg_class where relfilenode = '59258';
relname | relkind
---------------+---------
smallgroupset | r`
I can pg_dump the all the rows of the table smallgroupset and there is no data loss.
pg_dump -d tabula -t smallgroupset -f smallgroupset.sql -v > smallgroupset.log 2>&1
I've tried to reindex the table ( and the whole database) but it made no difference.
Can anyone advise the root cause of the errors?
Any help will be much appreciated.
orclp
Hello, On 4/15/24 11:30 AM, Oracle People wrote: > We have a few Postgres 16.2 servers running on VMs (RHEL 9.3) created on > VMware vSphere (7.0.3). 3PAR LUNs are used as their storages. > > On a couple of the PostgreSQL servers, there are errors such as: > > ERROR: could not read block 1264 in file "base/16520/59258": read only > 0 of 8192 bytes > > The errors occured when trying to insert rows to the tables. It looks like the same bug as https://www.postgresql.org/message-id/flat/1878547.tdWV9SEqCh%40aivenlaptop Maybe you can try to rebuild FSM : https://wiki.postgresql.org/wiki/Free_Space_Map_Problems#Cleanup_of_broken_FSM_files FYI, a fix has been committed : https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=93582974315174d544592185d797a2b44696d1e5 Regards, -- Adrien NAYRAT
Thanks Adrien, That bug does seem to fit what I'm experiencing. The block numbers in my error messages are just past the end of the relations. The fix was committed two days ago. Will it be included in the next minor release which is scheduled for May 9th? Best regards orclp On Mon, 15 Apr 2024 at 13:20, Adrien Nayrat <adrien.nayrat@anayrat.info> wrote: > > Hello, > > On 4/15/24 11:30 AM, Oracle People wrote: > > We have a few Postgres 16.2 servers running on VMs (RHEL 9.3) created on > > VMware vSphere (7.0.3). 3PAR LUNs are used as their storages. > > > > On a couple of the PostgreSQL servers, there are errors such as: > > > > ERROR: could not read block 1264 in file "base/16520/59258": read only > > 0 of 8192 bytes > > > > The errors occured when trying to insert rows to the tables. > > It looks like the same bug as > https://www.postgresql.org/message-id/flat/1878547.tdWV9SEqCh%40aivenlaptop > > Maybe you can try to rebuild FSM : > https://wiki.postgresql.org/wiki/Free_Space_Map_Problems#Cleanup_of_broken_FSM_files > > FYI, a fix has been committed : > > https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=93582974315174d544592185d797a2b44696d1e5 > > > Regards, > > -- > Adrien NAYRAT > >
On Mon, Apr 15, 2024 at 1:48 PM Oracle People <oraclepeople@gmail.com> wrote:
The fix was committed two days ago.
Will it be included in the next
minor release which is scheduled for May 9th?
Yes. Freeze date for a minor release is like a week prior to the public release date.
David J.