Undefined track indices in secondary vertex

I found a weird behaviour of the secondary vertices (SV). For some of them, the indices of the tracks are some ridiculous number while other parameters (while small) are defined. I don't know if it's a bug per se, but I think every SV should have a set of properly identified tracks. So maybe one solution would be to remove the SV if the tracks are undefined. Here are some examples coming from the minimum bias sample upgrade_mc_minbias_scifi_v5_000.mdf. I show 5 examples of badly defined SV and 1 example of a well defined SV.

Event number	SV index	trk1	trk2	p1	p2	doca
4319	78	1033490116	1016631570	0.098722	0.091006	0.080648
4343	33	1050544758	1036842758	0.163971	0.095338	0.084931
4405	47	1012691766	1093412331	0.029870	7.105389	0.105299
4444	0	1041083825	1026043605	4.852075	0.033044	0.011886
4677	5	995549666	998711489	0.165731	6.149944	5.206242
0	0	6	21	20339.230469	31352.421875	1.511394