To try to prevent further problems with /pnfs, we will install extra memory in several storage nodes.
To do this, the machines need to be shut off. It also means that during that time, the files residing on that machine will not be available.
In total, 5 machines will be restarted one by one (to keep the number of unavailable files to the strict minimum.
All affected storage nodes have had they RAM doubled, services have been restarted and are now fully in production. We expect the recent issues with /pnfs to be solved with this intervention.