The way an electromagnetic lock works (mag lock) is it stays locked when power is being supplied to it. If you remove power it is then unlocked or released. So if there is a case in which the lock is not releasing it is not the lock that is the issue, because there is not a way for the lock to remain locked if the power is indeed being removed from it which would be the function of the gate opener and a relay. So knowing this it should be looked at how the power is being disconnected from the lock at the beginning of the cycle. Some gate openers do not have a lock release so a relay is used. If a relay is being used the relay should be getting triggered by the motor leads. This is the more reliable way to release the lock. It would mean that if the motor is moving we would know for a fact power is being supplied to the coil side of the relay, it would be impossible for it not to be if the motor is moving. If this is the case with your opener the relay is probably bad. Or you are triggered the relay from someplace else on the board and you should move it. If you are using the board itself to interrupt power to the mag lock then you would need to see what conditions must be in place for the relay to activate and account for that. For example, on an Apollo 1050 board there is a mag lock release built in. The release will happen if the board is on an opening cycle from the closed limit switch. However if you have an issue where sometimes the closed limit switch is being reached in the closed position and sometimes the arm is stopping just shy of the limit switch (usually due to a physical limitation like a positive stop and the limit is set up to have the gate push against the stop to reach the switch), on the occasions where the limit switch is not reached the gate will appear closed because the lock is holding it there; but when triggered to open the lock will not release because it is not an opening cycle from the closed limit switch. After the motor tries to open and is prevented by the lock it will reverse directions and push harder on the closed stop, after a few of these obstructions it may then finally reach the limit switch in the closed position and then it will seem to open as normal.