After much tracking down I believe I narrowed the problem down to an
assumption of contiguous indices in table_data.c.
Any getnext request on the last row of a column followed by a gap would
produce nonsense. This patch seems to fix the problem:
--- table_data.c.orig Tue Oct 6 12:33:16
I've experienced an issue with 5.4.2.1 where my table has non-contiguous
column ids, and both snmpwalk and snmptable seem to stop at the gap.
i.e. (column ids 2,3,10)
snmptable -v2c -Ci -c public localhost:1616 mxOtherTable
SNMP table: MPATHIX-MIB::mxOtherTable
index
I tried using the valid_columns struct, didn't change anything:
netsnmp_table_set_multi_add_default_row( table,
2, ASN_OCTET_STR, 0, NULL, 0,
3, ASN_INTEGER, 0, defaultState, sizeof( defaultState ),
20, ASN_OCTET_STR, 0, NULL, 0,