too few non-missing observations, when perform ARIMA in R

I need perform forecat for two factors- STORE and ART.
So that I did

mydat=read.csv("C:/Users/synthex/Desktop/sales.csv", sep=";",dec=",")
df.sales.t <- acast(mydat, DAY ~ ART ~ STORE , value.var="SALES", fill=0)


[The structure of my data]

540.956 it's mean ART=№540 and STORE=№956

This loop for all variables

my_forecast <- function(x){
model <- arima(x, order = c(1, 1, 1))
fcast <- forecast(model, 1)

prog=lapply(df.sales.t[1:1], my_forecast)

Then I got the error

Error in arima(x, order = c(1, 1, 1)) : too few non-missing observations

What's wrong?