Context. Puppis A is a medium-age supernova remnant (SNR), which is visible as a very bright extended X-ray source. While numerous studies have investigated individual features of the SNR, at this time, no comprehensive study of the entirety of its X-ray emission exists. Aims. Using field-scan data acquired by the SRG/eROSITA telescope during its calibration and performance verification phase, we aim to investigate the physical conditions of shocked plasma and the distribution of elements throughout Puppis A. In doing so, we take advantage of the uniform target coverage, excellent statistics, and decent spatial and spectral resolution of our data set. Methods. Using broad-and narrow-band imaging, we investigate the large-scale distribution of absorption and plasma temperature as well as typical emission lines. This approach is complemented by spatially resolved spectral analysis of the shocked plasma in Puppis A, for which we divide the SNR into around 700 distinct regions, resulting in maps of key physical quantities over its extent. Results. We find a strong peak of foreground absorption in the southwest quadrant, which in conjunction with high temperatures at the northeast rim creates the well-known strip of hard emission crossing Puppis A. Furthermore, using the observed distribution of ionization ages, we attempt to reconstruct the age of the shock in the individual regions. We find a quite recent shock interaction for the prominent northeast filament and ejecta knot, as well as for the outer edge of the bright eastern knot. Finally, elemental abundance maps reveal only a single clear enhancement of the plasma with ejecta material, consistent with a previously reported knot, and no obvious ejecta enrichment in the remainder of the SNR. Within this region, we confirm the spatial separation of siliconrich ejecta from those dominated by lighter elements. The apparent elemental composition of this ejecta-rich region would imply an unrealistically large silicon-to-oxygen ratio for a core-collapse supernova. In reality, both the observed ejecta composition and their apparent distribution may be biased by the unknown location and strength of the reverse shock.