We have seen assert failures in debug builds in in-house testing on 1-CPU systems
(Raspberry Pi Zero and Beagblebone Black) in various tests occasionally.
The assert that fails is always the following.
%YDB-F-ASSERT, Assert failed in /Distrib/YottaDB/V999_R120/sr_port/relqueopi.c line 205
for expression (FALSE)
This is in the remqhi1() function when it has waited for 50 seconds to get the queue
header latch (GET_SWAPLOCK).
The 1-CPU systems that we have seen this failure on are slow boxes so this commit gives
such systems 4x the time (4 * 50 seconds) before signaling an assert failure.
In production builds, one would have gotten a DBCCERR error though we have not yet seen such
a failure in our in-house testing.
With the fix, the assert failure and DBCCERR error failures are expected to go away.
We have seen assert failures in debug builds in in-house testing on 1-CPU systems (Raspberry Pi Zero and Beagblebone Black) in various tests occasionally.
The assert that fails is always the following.
%YDB-F-ASSERT, Assert failed in /Distrib/YottaDB/V999_R120/sr_port/relqueopi.c line 205 for expression (FALSE)
This is in the remqhi1() function when it has waited for 50 seconds to get the queue header latch (GET_SWAPLOCK).
The 1-CPU systems that we have seen this failure on are slow boxes so this commit gives such systems 4x the time (4 * 50 seconds) before signaling an assert failure.
In production builds, one would have gotten a DBCCERR error though we have not yet seen such a failure in our in-house testing.
With the fix, the assert failure and DBCCERR error failures are expected to go away.