Convergence fails after adding weights in glmer

Question

Unfortunately without a reproducible example: I have a dataframe with 480,000 observations. Because of the nested structure of the data, I use a hierarchical model relying on the lme4 package. If I run a linear model there are no errors. However, because I have a dichotomous dependent variable, I use a generalized linear model (glmer). This also works fine, unless I add weights, which results in the following warning message:

Warning messages:

1: In eval(family$initialize, rho) : non-integer #successes in a binomial glm!

2: In (function (fn, par, lower = rep.int(-Inf, n), upper = rep.int(Inf, : failure to converge in 10000 evaluations

3: In optwrap(optimizer, devfun, start, rho$lower, control = control, : convergence code 4 from Nelder_Mead: failure to converge in 10000 evaluations

4: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.0608441 (tol = 0.002, component 1)

Here is the set-up:

f1 <- formula(dummy_dv ~ scale(log(count_iv+1)) + scale(sex) + scale(age) + year
          +(1 | country ))

model <- glmer(f1, data = dataset, family = binomial(link="logit"), weights = weight)

Year is a factor variable. I have also tried adjusting the optimizer and iterations:

model <- glmer(f1, 
              data = dataset, 
              family = binomial(link="logit"), 
              weights = weight, 
              control = glmerControl(optimizer="bobyqa", optCtrl = list(maxfun = 20000)))

This results in the following warning message:

Warning messages:

1: In eval(family$initialize, rho) : non-integer #successes in a binomial glm!

2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00660654 (tol = 0.002, component 1)

What can I do next? I was told that it is possible to use the results from the linear model as starting values for the glmer model. But I failed to implement this correctly.

Thank you for your help!