I, Christian Ehrhardt, hereby apply for per package upload rights to multipath-tools to work on it from the server-teams point of view (it is primarily owned by core-dev/foundations and cyphermox in particular).
Who I am
I'm member of Canonical in the Server Team since late 2015 and working there primarily on virtualization and ovisously server related tasks. I worked primiarly on qemu, libvirt and DPDK packaging and testing as well as many other minor packages that fall under the watch of the server Team (merges and SRUs). I also participate in Cloud-init/Curtin development as needed. I have a long history in System z regarding Linux Performance Analysis and due to that also work on s390x or performance related issues every now and then. In "the other life" I'm a 36 years old rather tall guy with the default family model: wife, son, daughter. And as all of us struggling to keep my hobbies of biking and playing saxophone alive in between work and family.
In "the other life" I'm a 36 years old rather tall guy with the default family model: wife, son, daughter. And as all of us struggling to keep my hobbies of biking and playing saxophone alive in between work and family.
My Ubuntu story leading to multipath-tools PPU
My Linux story starts with Gentoo around 2003. It suited my nature of optimizing things and emerge was a great package management tool at the time. Since around 2008 I changed all my systems to Ubuntu as it was just easier to consume in so many ways. Over time the full range from small embedded or HTPC to reasonable Server installations as well as on many of my Desktops were based on Ubuntu. I was closer to the Kernel for a while with a Thesis about I/O Schedulers in 2004. Then I worked with Linux as a performance analyst in IBM since 2004, but that was focused on System z where for a long time only Suse and RedHat existed. Yet that brought me into contact with so many benchmark, packages and the Kernel. A typical jack of all trades - expert in none but skilled in everything approach - causing a few fixes here and there around the open source community. I was switching between Performance analyst and a KVM Developer Role in between to further bolster my set of experiences. When the chance to join Canonical showed up in 2015 I saw the opportunity to combine my IBM Linux and virtualization history with really working on free software. Since then I'm slowly growing into more and more packages that fall under the server-teams responsibility. Multipath-tools is a special case as there are a few but still existing non-server use cases. Especially kpartx which is built from its sources is used in many other places of Ubuntu. Still from the common sense most use cases around the main part of multipath tools are
Examples of my work / Things I'm proud of
I'm a German, we are usually not proud but criticize ourself, trying to list some stuff still
- Optimizing KVM on s390x to outperform the classic z/VM Hipervisor in various areas up to 5x - full range from tuning to kernel patches.
- Identifying several complex performance issues over the years from faulty chip design up to sub-optimal cross application interactions.
recently making DPDK available on Ubuntu in (IMHO) the first user consumable way (https://insights.ubuntu.com/2016/05/05/the-new-simplicity-to-consume-dpdk/, unfortunately this upstream has surprise every release cycle but that forces me to learn which is good)
founding and driving joint packaging group to continue evolving DPDK in a colaborating way for Debian and Ubuntu (https://wiki.fd.io/view/Deb_dpdk)
Taking over most of qemu/libvirt work after Serge Hallyn left us which so far went well. Along that I was establishing a lot of extra testing to finally get rid of so many migration issues we had. Also that gives us a better warning system via jenkins
- ah the topic was "proud", every-time I can drop delta and actually can prepare and submit more to upstream and Debian to do more next time I'm "a bit" proud as I think that helps all the community sense as well as maintenance of those deltas (if only there would be more time for this sometimes)
Areas of work
Since joining Canonical I was part of the Server Team and are:
- working on Curtin and Cloud-Init with Scott Moser
- working on Subiquity with Ryan Harper
- working on various s390x issues with Dann Frazier and Dimitri Ledkov
- taking over DPDK completely from Stefan Bader
- working on openvswitch-dpdk with James Page
- taking over Qemu/Libvirt from Serge Hallyn
- working on many merges/fixes with James Page, Chris Arges, Martin Pitt, Adam Conrad, Robie Basak, ...
- regular bug triage duty for the server team helps to see what people really run into (and makes them feel cared about)
Things I could do better
There are still a lot of hidden "this is the way it is done in Ubuntu" gems that I have to uncover and adapt. But it gets better with every bug/upload/discussion I work on. I got the feeling recently that the last bunch of merges went by far smoother and easier. Still every now and then I run into something rather arcane or undocumented and un-earthing that can be hard.
Other than that I really want to be even more active in "helping people" - I started to keep an eye on IRC channels and askubuntu for that and it feels good - I just think that makes Ubuntu better overall.
Plans for the future
IMHO Linux in general has a very bad swapping spike behavior when passing into memory shortage that I really hate. I have quite some ideas about a triple watermark design and low prio async writers and all that stuff that I want to work on one day. That is a long term thing I still carry from my performance times and I don't know if I can ever tackle this - but it is one particular thing in the back of my mind. Well I said that last time I applied for some upload rights and nothing moved so I assume that just will be the case for a while :-/
Well I said that last time I applied for some upload rights and nothing moved so I assume that just will be the case for a while :-/
What I like least in Ubuntu
I cut my old deail a bit short - TL;DR I think we are nod good enough on QA and SRUs. I think we could double the efforts on those and still not work in-vain or have anybody idling.
If you'd like to comment, but are not the applicant or a sponsor, do it here. Don't forget to sign with @SIG@.
I want to refer to my core-dev / server-dev application endorsements as a base.
I've reviewed a few change sets from Christian related to multipath-tools, and although I can only find one instance of an upload claimed under his name, I'm very confident that he has the necessary knowledge and skill to properly handle directly uploading multipath-tools by himself. Christian understands well the impact of multipath-tools changes on our users.
Specific Experiences of working together
We've interacted a lot on multipath-tools, including reviewing for the upload of https://launchpad.net/ubuntu/+source/multipath-tools/0.6.4-3ubuntu1, which he did by himself (multipath-tools was under the server team packageset for a while, and thus he could do the upload on his own). multipath-tools has since been moved back to core, which removed it from the list of packages Christian could upload. I believe he should still have this access, although we want the slightly higher amount of scrutiny / apparent importance conveyed by the package being in the core packageset.
Areas of Improvement
The only thing I could say for Christian in terms of area of improvement would be to not wait until an application has my endorsement before adding himself to the DMB's agenda -- this translates to a "loss of time" where he could have made bug fixes directly to multipath-tools. It also translates into something that can be applied to uploads in general: things might not be quite perfect, but if they are a net improvement on what is currently in the archive, sometimes it's still a good idea to upload now rather than later. Obviously this is something to review on a case by case basis, but inactivity due to a quest for perfection can also cost us worthwhile improvements.
== <SPONSORS NAME> == === General feedback === ## Please fill us in on your shared experience. (How many packages did you sponsor? How would you judge the quality? How would you describe the improvements? Do you trust the applicant?) === Specific Experiences of working together === ''Please add good examples of your work together, but also cases that could have handled better.'' ## Full list of sponsored packages can be generated here: ## http://ubuntu-dev.alioth.debian.org/cgi-bin/ubuntu-sponsorships.cgi? === Areas of Improvement ===