Looking at ER8 and O1 data with the long transient search pipeline stampas we have noticed an excess of triggers corresponding to 2 narrow frequency lines at 39Hz and 43 Hz (plot 1). This excess of triggers happen at the beginning of H1 relock periods and is so large that it is visible in the spectrogram of h(t) (plot2). After a dozen of minutes, the excess of loud triggers disappear along with the lines. We are seen this problem since the beginning of ER8 but actually it seems that since October 8th, the excitation is smaller at each relock. Plot 3 shows that the long transient search is dominated by these triggers. We will develop a veto against these triggers, but we would be happy to see the pb fixed in the data!