Data on xenoliths carried by lavas of Avacha volcano and ejected during eruptions are used in discussing the partial melting of metasomatized rocks of the lithospheric mantle (which were tectonically reformed by seismic events beneath the frontal zone of Avacha volcano) and the associated growth of minerals from a gas phase in open fractures above the magma chamber feeding the volcano. Based on a comprehensive study of representative mantle xenoliths from explosions of the volcano, we developed and numerically studied a mathematical model of convective heat and mass transfer in the permeable zone above the feeding magma chamber. The model enables analysis of the area of active seismic destruction of rocks at depths of 30–70 km beneath Avacha volcano. This analysis involves (1) current qualitative genetic hypotheses concerning the nature of the identified textural and structural relationships between the xenoliths and lavas transporting them, (2) data on the composition of mineral-hosted fluid and melt inclusions, (3) estimated homogenization temperatures of the inclusions, and (4) hypothetical melting mechanisms of metasomatized ultramafics and the associated growth mechanism of minerals in open fractures. It is demonstrated that fractured porous seismogenic regions above magma chambers are coupled with a set of convective processes of fluid-mediated heterophase heat and mass transfer in the lithosphere: infiltration metasomatism at the spinel depth facies and the sublimation and condensation of major components from a gas phase. It is hypothesized that local partial decompression melting of highly heated ultramafic and mafic rocks may be followed by major seismic events. It is shown that spinel-facies infiltration metasomatism in fractured mantle rocks may be associated with the growth of crystal crusts and clusters of clinopyroxene and amphibole from a gas phase in open fractures. There are two paths by which decompression partial melting takes place: (1) the melting of spinel aggregates with the host crystals of “metasomatic” orthopyroxene, which replaced olivine in harzburgites, and (2) the development of clinopyroxene, amphibole, and relict orthopyroxene vein fillings. To test these hypotheses, physical experiments were conducted with the use of convective heating of samples of natural xenoliths. The experiments were carried out in a flow-through reactor capable of reproducing the partial melting phenomena and in an electron gun with a power unit for a welding system with a dense electron beam. The latter experiments simulated the sublimation and deposition of major components from a gas phase over the solidus boundary for all petrographic types of mantle ultramafic rocks. Neither structural nor mineralogical indications of the infiltration of magmatic fluid were detected in the experiments, and no traces of diffusion–reaction zoning of the “melt–rock” type were found in the xenoliths.