Dropping and creation of tables frequently may be corrupting other tables

Hi,

We have a function which uses dblink to connect to grid and runs some join queries.
In that function we also have a creation and dropping of lots of tables.
WE have loaded data into some tables.Evrything was fine.But after running the function we are getting problems with already loaded tables :
SQLException: ERROR: Failed To Get Results For ( SQL , NodeURL) : ( SELECT count(*) AS "XCOL1","captor_tdr_msc_b_100210"."flexible_field_20" AS "XCOL2" FROM "captor_tdr_msc_b_100210" group by "captor_tdr_msc_b_100210"."flexible_field_20") eQS Node 9 has aborted execution, cause is: java.sql.SQLException : ERROR: could not read block 380578 of relation 3054709/1285040/3079111: read only 4096 of 8192 bytes

If it is a hardware issue this problem should occur immediately after loading.But we were able to generate reports .Only after the function execution we are getting errors of this kind.

Thanks
Sreelatha

I am sorry, but GridSQL relies on the underlying databases for storage, and those errors you see are read errors that the RDBMS encounters from the storage system.

I think you reported similar errors in the past. Did they happen only when using dblink, or at other times? If your hardware is flakey, it is possible that it would work for a while, then encounter an error.

Also, do you only see this error for one of the physical servers? Is it always for the same disk array/controller, or all? If it happens for just one RAID array, it probably points to a problem there. If it always happens for multiple storage devices connected to one particular controller, it is probably the controller.

This seems to be an ongoing issue. Can you please try and see if you can isolate to see if it happens for only a certain hardware component? I would also check the disk controller vendor's website and see if there have been reported issues for your model and firmware and drivers, and consider

Regards,

Mason