Comment 5 for bug 1584052

Revision history for this message
Oliver O. (oliver-o456i) wrote :

The Xenial test kernel with a pick of 4da2e26 resolves the bug, no negative side effects were encountered.

With the Xenial release kernel, there were 18 MySQL server crashes in a 24 hour period due to false ENOSPC indications. With the test kernel, not a single crash occured in a 24 hour period.

Additionally, I've checked with a script derived from https://patchwork.kernel.org/patch/7967161/ (see attachment), which creates a temporary btrfs test image in a plain file. The script provoked ENOSPC errors with the release kernel on most runs (sometimes it missed, presumably due to timing conditions). The script could not provoke a single ENOSPC condition with the test kernel.