So, if that is the case, could you check the RACH root sequence?
It could be that no shift bits is possible.
Can you please tell a little bit more?
What do I have to check?
There should be a parameter on the gNodeB for the Zadoff-Chu sequence. It should be called something like rootSequence.
So the RACH is a UE signature to identify itself to gNodeB, from SIB1 UE knows the RACH format used and based on that knows how many sequences are there in the RACH.
The bits used in each sequence is 139. So each UE do a shift for a specific number of bits for a certain sequence to send to gNodeB.
From memory, a single sequence can be shifted 3 times only for 40MHz BW “there is a calculation for it” and since you must have 16 signature in one cell, you need to allocate 6 root sequence for each cell (rounded from 5.55)
We have long preamble sequence which is 839.
From SIB1 we have zeroCorrelationZoneConfig=10 => gives cycle shift 76 and root sequence per logical root sequence =11 and 6 root sequence per cell.
From prach-RootSequenceIndex l839 : 456=> logical root sequence is 456 which points toward physical root sequence 230,609,232,607,266,577
Now for above these 6 physical root sequence with 11 shift UE can generate 64 preambles easily.
This is my understanding as per configuration so why UE won’t shift bit to generate preamble?
Maybe it is from higher layer, if UE attached to network and then receive specific fail, then it will not send new request.
Can you check QXDM logs just after this?
Can you also send SIB1 parameters related to RACH?
Shared RACH parameters above.
Just after this UE does reselection to different cell and completed RACH and made RRC.
SIB1 RACH parameters: 
Can you increase the preambleReceivedTargetPower from -116 to say -106 or -96 and try once.
At gNB side -116 seems very low.
How many receiver antennas you are using?
What is your TX power?
Can request for that.
Yes the value i also observed less compare to other cells but my point was, UE was able to complete RACH just before this fail on same cell.
So what happened with the UE to not transmit msg1 this time.
UE location was same, we did it in stationary location.
Even with the same location, RSRP can fluctuate.
Yes but I believe at least UE should attempt with Pcmax even in high path loss as per Prach power calculation.
But we haven’t observed any attempt for msg1.
It might be UE issue we are going to raise it with OEM.