An attempt is made to resolve the controversy related to the low temperature phase (ground state) of the low-doped ferromagnetic (FM)-insulator(I) manganite through bulk magnetic measurements on La0.9Sr0.1MnO3 sample. It is shown that the FM phase, formed out of well defined transition in the low-doped system, becomes inhomogeneous with decrease in temperature. This inhomogeniety is considered to be an outcome of the formation of orbital domain state of eg-electrons having hole rich (metallic) walls separating the hole deficient (insulating) regions. The resulting complexity brings in metastability and glassy behaviour within the FM phase at low temperature, however, with no resemblance to spin glass, cluster glass or reentrant phases. It shows ageing effect without memory but magnetic relaxation shows signatures of inter-cluster interaction. The energy landscape picture of this glassy phase is described in terms of hierarchical model. Further, it is shown that this inhomogeneity disappear in La0.9Sr0.1MnO3.08 where, the orbital domain state is destroyed by self doping resulting in reduction of Mn 3+ and hence eg-electrons. The ferromagnetic phase of the nonstoichiometric sample, does not show glassy behaviour. It neither follows 'hierarchical model' nor 'droplet model' generally used to explain glassy or inhomogeneous systems. Its magnetic response can be explained simply from the domain wall dynamics of otherwise homogeneous ferromagnet.