Is it acceptable to fix convergence errors by restarting the model from the failure point

I'm running a GLMM but I keep getting convergence errors (failure to converge).

I was able to get around the messages by putting the code:

ss <- getME(Model5,c("theta","fixef"))
ss
Model5b <- update(Model5,start=ss)

Is this an OK thing to do or is it considered 'cheating'?