The mass of galaxy clusters can be inferred from the temperature of their X-ray emitting gas, $T_{mathrm{X}}$. Their masses may be underestimated if it is assumed that the gas is in hydrostatic equilibrium, by an amount $b^{mathrm{hyd}}sim(20pm10)$ %
suggested by simulations. We have previously found consistency between a sample of observed textit{Chandra} X-ray masses and independent weak lensing measurements. Unfortunately, uncertainties in the instrumental calibration of {em Chandra} and {em XMM-Newton} observatories mean that they measure different temperatures for the same gas. In this paper, we translate that relative instrumental bias into mass bias, and infer that textit{XMM-Newton} masses of $sim 10^{14},mbox{M}_{odot}$ ($> 5cdot 10^{14} mbox{M}_{odot}$) clusters are unbiased ($sim 35$ % lower) compared to WL masses. For massive clusters, textit{Chandra}s calibration may thus be more accurate. The opposite appears to be true at the low mass end. We observe the mass bias to increase with cluster mass, but presence of Eddington bias precludes firm conclusions at this stage. Nevertheless, the systematic textit{Chandra} -- textit{XMM-Newton} difference is important because {em Planck}s detections of massive clusters via the Sunyaev-Zeldovich (SZ) effect are calibrated via {em XMM-Newton} observations. The number of detected SZ clusters are inconsistent with {em Planck}s cosmological measurements of the primary Cosmic Microwave Background (CMB). Given the textit{Planck} cluster masses, if an (unlikely) uncorrected $sim 20$ % calibration bias existed, this tension would be eased, but not resolved.
