If you set the line level and calibrated the VU meters first, all you have to worry about is what the meters indicate. Some calibration procedures will say the test tape needs to show 0VU at a certain voltage from the output, but it’s redundant if you’ve already calibrated the meters. It’s really a double check in case you missed something in a previous step.
The only thing you can do if the meter and output voltage don’t agree while playing the test tape is to go back to the previous line level/meter calibration steps using the 1kHz test tone from the oscillator.
You’ll be able to verify this later after the record calibration step using the blank SM911. When you record a 1kHz tone @ 0VU it will play back @ 0VU and your output will read 1.23 volts for +4 dBm line level.
If the instructions for the 8-track are anything like the MX5050 half-track, it is a bit confusing. The order is screwy. It’s easiest to follow the order below once mechanical adjustments are made:
1. Line level input/output matching, including zeroing meters.
2. Reproduce level adjustment - playing 1kHz tone on test tape, reference VU meters
3. Reproduce equalization
4. Record bias level
5. Record level using SM911 or tape you'll be using most
6. Record equalization
Your track-2 problem could be bias related or a relay problem, which is fairly common with the MX5050... or something else. It's not likely the head if all other tracks/channels seem ok.