Second SIP call between two CS1000s fails

[twlcolorado]

I have a customer with a CS1000M Single Group (Option 61C) and a CS1000E chassis, both on 7.6. There is a T-1 between the two locations that is being used as a data pipe. On each end is a SR10004. This has been in place and working for years. They are using Session Manager for network routing and registration.

Recently they started noticing that some calls are failing. It was determined that one call will go through, but no others. A d-channel trace shows the following:

DCH 12 OMSG SETUP REF 00000197 CH 252 0 0 22 TOD 9:06:37
FEAT :CDS
FEAT :NCID
PROGRESS: ORIG ADDR IS NOT ISDN
CALLING #:XXXX NUM PLAN: PRIVATE/ABBREVIATED (CDP)
CALLED #:101XXXX NUM PLAN: PRIVATE/ABBREVIATED (CDP)

DCH 12 IMSG CALLPROC REF 00000197 CH 252 0 0 22 TOD 9:06:37

DCH 12 IMSG DISC REF 00000197 CH 252 0 0 22 TOD 9:06:37
CAUSE :NTWK TEMP FAIL

DCH 12 OMSG RELEASE REF 00000197 CH 252 0 0 22 TOD 9:06:37

DCH 12 IMSG REL COMP REF 00000197 CH 252 0 0 22 TOD 9:06:37

That data pipe is used by the customer for other things, and there have been no issues with that.

At the CS1000E end the SR1000, Call Server, UCM Server and the data switches have all been rebooted with no effect.

Any ideas?