Cluster index oddities

[Joseph Crosby]

Good day,

A few days ago we had an app issue that yelled:
2017-Nov-10 01:16:08.237019: Got temporary error 785 ‘Schema object is busy with another schema transaction’ from NDBCLUSTER

While investigating I discovered the only thing happening in the cluster at that time was the following:
ndb_50_cluster.log.2:2017-11-10 01:16:06 [MgmtSrvr] INFO -- Node 1: index-build table 16 index: 67 processed 0 rows
ndb_50_cluster.log.2:2017-11-10 01:16:06 [MgmtSrvr] INFO -- Node 2: index-build table 16 index: 67 processed 0 rows
ndb_50_cluster.log.2:2017-11-10 01:16:06 [MgmtSrvr] INFO -- Node 3: index-build table 16 index: 67 processed 0 rows
ndb_50_cluster.log.2:2017-11-10 01:16:06 [MgmtSrvr] INFO -- Node 4: index-build table 16 index: 67 processed 0 rows

These entries run until:

ndb_50_cluster.log.2:2017-11-10 01:16:19 [MgmtSrvr] INFO -- Node 1: index-build table 240 index: 283 processed 0 rows
ndb_50_cluster.log.2:2017-11-10 01:16:19 [MgmtSrvr] INFO -- Node 2: index-build table 240 index: 283 processed 0 rows
ndb_50_cluster.log.2:2017-11-10 01:16:19 [MgmtSrvr] INFO -- Node 3: index-build table 240 index: 283 processed 0 rows
ndb_50_cluster.log.2:2017-11-10 01:16:19 [MgmtSrvr] INFO -- Node 4: index-build table 240 index: 283 processed 0 rows



My questions seem to be:

1) Does index-build relate to the rebuilding of indexes? I can't find anything online.

2) And if it does why would NDB decided to rebuild them on its own? I cannot find any mention of this in a conf file or any user input.


I'm running on the assumption that something caused an index build of the cluster which in turn caused our app to fail out because the Database/Tables it was trying to access were busy. But some verification would be greatly appreciated.