db2start may hang and lead to group restart in pureScale because
of no proper timeout before calling TSA commands. When the
db2start command received the interrupt , we did a group restart
of all the nodes
We need to add a default timeout for the sqlhaCB structure
Here in sqleIssueStartStop we do not set the timeout before
calling sqlhaStopSDInfrastructure function
// try a cleanup if this is clusterEnv and db2start an
individual CA
// this is to reverse the RG lock/unlock that may have been
done
// by sqlhaStartSDInfrastructure above
// Do not do this cleanup if it was determined that the
db2start command
// was invalid as that implies the SD infrastructure is already
up.
if (clusterEnv &&
status.caNodenum &&
(SQLE_DB2ADMIN_DB2START == command) &&
(SQLE_RC_INVSTRT != remoteSqlca.sqlcode))
{
SQLHA_CONTROL_BLOCK sqlhaCB = {0x00};
SQLHA_CLUSTER_OPERATION_RESULT_LIST *pResultList = NULL;
int tempRc = 0;
tempRc = sqlhaStopSDInfrastructure(pInstance,
&pResultList,
&sqlhaCB,
pStartOptions->iNodeNum); |
