Currently we don't raise open fd limit for sandbox=chroot mode. And for sandbox=namespace mode we do raise it.
I would say try to match the logic we have in C version. Idea is to get a minimum default in all configurations so that it works reasonably well for all configurations. - If user passed option -o rlimit_nofile, simply follow that. - If user did not pass anything, then try to keep a default of 1M. This should work reasonably well for lot of workloads out of the box. - Call getrlimit() before raising limits. If caller has already setup the limit higher than 1M, don't do anything. --- https://gitlab.com/virtio-fs/virtiofsd/-/issues/37 _______________________________________________ Virtio-fs mailing list Virtio-fs@redhat.com https://listman.redhat.com/mailman/listinfo/virtio-fs
