NDSolve not working : Step size zero, singularity/stiff system suspected

Question

I am trying to solve a set of four coupled equations through NDSolve. However, I am getting the error "NDSolve::ndsz: step size is effectively zero; singularity or stiff system suspected.." I think this is a common error when there exists very large or very big numbers, but I can't see any such numbers in my system. So, I am worried what could be the reason ? The set of equations are given below :

eqn1 = M'[a] + 
Log[10]/Sqrt[1/3  (B[a] 10^(-3 a) + R [a] 10^(-4 a) )] (1/M[a]) ^2;
eqn2 = R'[a] + 10^a/M[a] M'[a] B[a];
eqn3 = B'[a] - 1/M[a] M'[a] B[a];
eqn4 =  s'[a] + 3 s[a] Log[10] + 
1/  (s[a]^((1/3))) B[a]/(M[a] 10^a) M'[a];
sol1 = NDSolve[{eqn1 == 0 , eqn2 == 0, eqn3 == 0, 
eqn4 == 0, M[2] == 2, R[2] == 1, 
B[2] == 0.001, s[2] == 0.1}, {M, R, B , s}, {a, 2, 8}]

Answer 1

Seems for all positive initial conditions (m2) for M, system has a singularity. But for negative m2, you get a full a-range solution.

ndsol[b2_, m2_, r2_, s2_] := 
  First@NDSolve[{eqn1 == 0, eqn2 == 0, eqn3 == 0, eqn4 == 0, 
B[2] == b2, M[2] == m2, R[2] == r2, s[2] == s2}, {B, M, R, s}, {a,
 2, 8}]

Plot[Evaluate[M[a] /. ndsol[.001, 2, 1, .1]], {a, 2, 
   2 + .9999*(puf[[1, 2, 1, 1, 2]] - 2)}]

Manipulate[
  Plot[Evaluate[M[a] /. (puf = ndsol[b2, m2, r2, s2] // Quiet)], {a, 2,
2 + .9999*(puf[[1, 2, 1, 1, 2]] - 2)}], {{b2, .001}, -3, 3, 
Appearance -> "Labeled"}, {{m2, -2}, -5, 5, 
Appearance -> "Labeled"}, {{r2, 1}, -5, 5, 
Appearance -> "Labeled"}, {{s2, .001}, -1, 1, 
Appearance -> "Labeled"}]