How can I use idh and random when I use hupossion in mcmcglmm?

Question

Here is my problem: I need to use hupossion in MCMCglmm package. Here is my prior:

> prior = list(R = list(V = diag(2), nu = 0.002),
 +                G = list(G1 = list(V = diag(10), nu = 0.002),
 +                           G2 = list(V = 1, nu = 0.002)))

and here is my training:

> myTrain <- MCMCglmm(pipenumyearnosameday ~ trait - 1 +
+                         at.level(trait,1):lastyearnumnosameday + 
+                         at.level(trait,1):pipeInsideSizemm,
+                     random = ~idh(pipematerialgroup1):installyeargroup1,
+                     data = train,
+                     prior = prior,
+                     verbose = F,
+                     pr=T,
+                     family = "hupoisson",
+                     rcov = ~idh(trait):units)

There are 5 levels in the attribute pipematerialroup1. I know sth must be wrong with the prior. I got the following information:

Error in MCMCglmm(pipenumyearnosameday ~ trait - 1 + at.level(trait, 1):lastyearnumnosameday + : prior$G has the wrong number of structures

there is one more thing I need to say:

myPredict <- predict(myTrain,newdata = myTest,type = "response",marginal = ~idh(trait):pipematerialgroup1,interval = "confidence") 

This is my prediction part when I just take pipematerialgroup1 as random effect. If combine pipematerialgroup1 and installyeargroup1, what this prediction would be?

Can someone help me get the prior right? Thanks a lot.