Thanks to everyone who participated at this month's meeting, and especially to Mark Maybee for running the meeting and to Serapheim for taking notes.
video recording: https://youtu.be/5Lby_Vl6H88 notes: - OpenZFS Dev Summit: Dates, planning, etc. - Registration for talk proposals opens on July 23rd and deadline is on August 30th - Conference Nov 8-9th - Remote or hybrid (in-person and remote) conference is being discussed. Would folks be interested in the hybrid option? Only Allan expressed interest. Most interest is in remote-only. - Run ztest with asan (Address Sanitizer malloc debugger) (Matt/Mark) There was some initial issue with ztest, issue 12215 that needs to be tracked down before we can enable ASan Related: MarkJ@FreeBSD also using KMSAN: https://github.com/openzfs/zfs/pull/12383 - Status update on Panzura’s temporal dedup? (matt:) looks like the last update was 4/28/2020, they have “de-prioritized” this upstreaming work No one on the call has update. - Allan Jude: - Linux Namespace support PR12263 is ready for review - About time to remove WIP status for vdev properties PR11711, new tests are being written. - Klara is working on zvol performance, if you have known pathologies, please share - Allan Jude brought up enabling LZ4 compression by default. - Ryan Moller suggests changing default for xattr to SA - Alexander Motin suggests changing default zvol block size from 8kb to 16kb - Discussed possibly forcing allocations to be far apart on the disk (ala ditto blocks) to improve flash performance - Xattr compat (Ryan): https://github.com/openzfs/zfs/pull/11919 Requesting design/code review. Tony to reach out to Brian. - Formalize a procedure for ‘erratas’, like the encryption+ project dnode issue, or even the zstd big-endian issue (like a bug@ not a feature@ flag) Due to zfs projects being ported after encryption, #7177 (include the project dnode as part of the objset hash on encrypted datasets) was missed (this appears to have happened with the macOS port as well). As a result, there are now encrypted zfs datasets on illumos that are incompatible with OpenZFS and vice versa. While it is possible to re-try any objset hash failures for compatibility between the two, ‘fixing’ such a dataset presents some thorny issues. One approach that was floated in the illumos community (see https://illumos.topicbox.com/groups/developer/T1981ce734a4a8c50-M4a3b3c0aff56de6ac0ceddfc) was to propose an SPA feature that’s largely a dummy feature on non-illumos platforms (i.e. is ignored), which would allow for explicit identification of the two pools. On Mon, Jul 19, 2021 at 1:41 PM Matthew Ahrens <[email protected]> wrote: > The next OpenZFS Leadership meeting will be held tomorrow, July 20, 1pm > Pacific time. > Everyone is welcome to attend and participate, and we will try to keep the > meeting on agenda and on time. The meetings will be held online via Zoom, > and recorded and posted to the website and YouTube after the meeting. > > For more information and details on how to attend, as well as notes and > video from the previous meeting, please see the agenda document: > > > https://docs.google.com/document/d/1w2jv2XVYFmBVvG1EGf-9A5HBVsjAYoLIFZAnWHhV-BM/edit > > --matt > ------------------------------------------ openzfs: openzfs-developer Permalink: https://openzfs.topicbox.com/groups/developer/Ted9f300b994b7c2a-Ma3007de6669d1c1748c937bd Delivery options: https://openzfs.topicbox.com/groups/developer/subscription
