Cannot find initial connected bush

[finnwiz]

I ran a simple test (test_net.txt and test_trips.txt), which worked and was great (and got the expected result)!

./bin/tap .1 1 test_net.txt test_trip.txt
Nodes, arcs, zones, thrunode: 4 6 2 0
Distance factor, toll factor: 0.000000 0.000000
4 nodes, 6 arcs, and 2 zones
File read and memory allocated.
Forward and reverse star lists generated.
Starting Algorithm B...
Warning: Creating 2 artifical arcs to ensure strong connectivity.
Finalizing network setup
Finalized network setup
Creating Initial bushes
Initializing Algorithm B
Initialized flows
Read matrix 0
Creating batch 0
Finished shortest path bush preds 0
Finished shortest path bush preds 1
Created batch 0
Initialization done in 0.000 s.
Iteration 1: gap 3.534204911719327, obj 248.451103865328, time 0.000, 4 shifts
Iteration 2: gap 0.771018213410878, obj 168.877623226108, time 0.000, 26 shifts
Iteration 3: gap 0.150833572377848, obj 152.910162001144, time 0.000, 48 shifts
Iteration 4: gap 0.020958208823608, obj 150.846825064805, time 0.000, 70 shifts

But when I run my "real" problem (using net.txt and trips.txt), I get an exciting error:

npadon@HO1-PW03D5WL:~/tap-b$ ./bin/tap .1 1 net.txt trips.txt
Nodes, arcs, zones, thrunode: 345843 650887 1504 0
Distance factor, toll factor: 0.000000 0.000000
345843 nodes, 650887 arcs, and 1504 zones
File read and memory allocated.
Forward and reverse star lists generated.
Starting Algorithm B...
Warning: Creating 14677 artifical arcs to ensure strong connectivity.
Finalizing network setup
Finalized network setup
Creating Initial bushes
Initializing Algorithm B
Initialized flows
Read matrix 0
Creating batch 0
Finished shortest path bush preds 0
Fatal error: Cannot find initial connected bush.

I tried with binaries from both make and make parallel. In the full log, I can also see that some of the bush prediction are -1 (ostensibly the value of IS_MISSING and the cost is infinite...). Additionally, I created a different version of net.txt that had fully bi-directional edges (some of the links in the attached net.txt are only one-way) - but got the same error.

I also confirmed that my graph (both the one in net.txt and the crazier one with bi-directional links) is strongly connected (using igraph) so would appreciate any pointers (even if general).

Thanks in advance.

net.txt.gz
test_net.txt
test_trip.txt
trips.txt
full_log.txt.gz