Remount sysfs to make its superblock rw: mount -o remount,rw /sys |

When testing Amazon EFS volumes in containers, the author discovered that NFS readahead performance varied unpredictably because containers sharing a network namespace share a single sysfs superblock. Since sysfs flag states are determined by the first mount in that network namespace, subsequent mounts cannot write to sysfs if the initial mount was read-only. The author explains that remounting sysfs using mount -o remount,rw /sys clears the read-only superblock flag idempotently and resolves the readahead failure. Open article →