Recent problems with any query


#1

We are running 14.8 on 34 nodes, with plenty of available physical memory.
At the time of the query, each of the 34 nodes had 25GB of free physycal memory.
There is plenty of disk space for storage. On each node, unshared 446GB used on a 13TB volume .

What would cause this?

[quote]AFL% list(‘arrays’);
SystemException in file: src/query/QueryProcessor.cpp function: validateQueryWithTimeout line: 361
Error id: scidb::SCIDB_SE_EXECUTION::SCIDB_LE_RESOURCE_BUSY
Error description: Error during query execution. Not enough resources: not enough resources to start a query.[/quote]

Thanks, y’all


#2

This error is raised because scidb suspected that there is a deadlock that would prevent the query from completion. The deadlock can occur due to thread starvation if multiple coordinators are in use. By default the timeout is set to 30 sec, you can use the ‘deadlock-timeout’ config.ini parameter to increase it. Another option is to increase the number of ‘execution-threads’ (but keep in mind that over-subscription may cause thrashing). In any case, SCIDB_LE_RESOURCE_BUSY is an expected error condition because at high enough load scidb will have to stop accepting additional work.