Over on the project page and on launchpad you can now download libeatmydata 105.
This release fixes a couple of bugs that came in via the Debian project, including a rather interesting one about some binaries not running .so ctors to properly init libeatmydata and the code path in the libeatmydata open() not really dealing with being called first in this situation.
Enjoy!
This is the Call for Papers for the Developer, Testing, Release and Continuous Integration Automation Miniconf at linux.conf.au 2015 in Auckland. See http://linux.conf.au
This miniconf is all about improving the way we produce, collaborate, test and release software.
We want to cover tools and techniques to improve the way we work together to produce higher quality software:
– code review tools and techniques (e.g. gerrit)
– continuous integration tools (e.g. jenkins)
– CI techniques (e.g. gated trunk, zuul)
– testing tools and techniques (e.g. subunit, fuzz testing tools)
– release tools and techniques: daily builds, interacting with distributions, ensuring you test the software that you ship.
– applying CI in your workplace/project
We’re looking for talks about open source technology *and* the human side of things.
Speakers at this miniconf must be registered for the main conference (although there are a limited number of miniconf only tickets available for miniconf speakers if required)
There will be a projector, and there is a possibility the talk will be recorded (depending on if the conference A/V is up and running) – if recorded, talks will be posted with the same place with the same CC license as main LCA talks are.
CFP is open until midnight November 21st 2014.
By submitting a presentation, you’re agreeing to the following:
I allow Linux Australia to record my talk.
I allow Linux Australia to release any recordings of my presentations, tutorials and minconfs under the Creative Commons Attribution-Share Alike License
I allow Linux Australia to release any other material (such as slides) from my presentations, tutorials and minconfs under the Creative Commons Attribution-Share Alike License.
I confirm that I have the authority to allow Linux Australia to release the above material. i.e., if your talk includes any information about your employer, or another persons copyrighted material, that person has given you authority to release this information.
Any questions? Contact me: stewart@flamingspork.com
There seems be a bit of an exodus from the MariaDB Foundation board recently… I’m not sure exactly what to make of it all, but the current members according to https://mariadb.org/en/foundation/ are:
With Jeremy Zawodny being the only non-MariaDB Corp member.
Recently, Jeremy Cole asked some people about their membership:
@jeremycole Nope. Resigned months ago.
— Mike Milinkovich (@mmilinkov) October 2, 2014
@jeremycole Hi Jeremy, no, sorry, I resigned a little while back.
— Andrew Katz (@andrewjskatz) October 1, 2014
@jeremycole yes, though I'm having some doubts about its future (for a variety of reasons) :-(
— Jeremy Zawodny (@jzawodn) October 1, 2014
I’m a little worried for the project, the idea of a foundation around it and for people I count as friends who work on MariaDB.
Good news everyone!
MySQL 5.7.5 is out with a bunch more patches for running well on POWER in the tree. I haven’t yet gone and tried it all out, but since I’m me, I look at bugs database and git/bzr history first.
On Intel CPUs, when you’re spinning on a spin lock, you’re meant to execute the PAUSE CPU instruction. This tells the CPU that other execution threads in the same core should be given priority as you are currently not doing anything productive. Without this, you’re likely going to hurt on hyperthreaded CPUs.
In MySQL, there are custom spinlocks in order to do interesting adaptive mutex things to attempt to squeeze the most performance possible out of modern systems.
One of the (not 100% ready, but close) bugs with patches I submitted against MySQL 5.7 was for using the equivalent of the PAUSE instruction for POWER CPUs. On POWER, we’re a bit different, you can actually set priorities of threads (which may matter more, as POWER8 CPUs can be in SMT8 mode – where there are *eight* executing threads per core).
So, the good news is that in MySQL 5.7.5, the magic instructions for setting thread priority are in! This should mean great things for performance on POWER systems with any of the SMT modes enabled.
The next interesting part of this is how it interacts with other KVM guests on a system. At least on POWER (and on x86 as well, although I won’t go into details here) there’s a hypervisor call that a guest can make saying “hey, I’m spinning here, perhaps you want to make sure other vcpus execute so that at some point I can continue”. On POWER, this is the H_CONFER hcall, where you can basically do a directed yield to another vcpu (the one that holds the lock you’re trying to get is a good idea).
Generally though, it’s only the guest kernel that does this, not userspace. You can see the H_CONFER call in __spin_yield(arch_spinlock_t*) and __rw_yield(arch_rwlock_t*) in arch/powerpc/lib/locks.c in the kernel.
It would be interesting to see what extra we could get out of a system running multiple guests with MySQL servers if InnoDB/MySQL could properly yield to the right vcpu (well, thread I guess).
Good news everyone! Tyan has announced the availability of their first OpenPOWER system! They call this a Customer Reference System, which means it’s an excellent machine to start poking at OpenPower and POWER8 (or deploying applications on).
Because it’s an OpenPower machine, it runs the open source Open Power firmware (all up on github) and will happily run Linux (feel free to port your other operating system kernels). I’ll be writing more on the OpenPower firmware soon as, well, technical details are fun!
Ubuntu 14.10 is listed as recommended as not only have they been building for POWER8 but have spent some time ensuring things work fairly well out-of-the-box (both as a KVM guest and running native on the bare metal). Or, you can always just boot whatever the mainline kernel is at – build for the POWERNV (POWER non-virtualized) platform (be sure to include all the required drivers) and have fun!
Good news for those wanting to run MariaDB on POWER systems, the latest 10.0 bzr tree (as of a couple of weeks ago) builds and runs well!
I recently pulled the latest MariaDB 10.0 from BZR and built it on a POWER8 system in the lab to run some quick tests. The MariaDB team has done some work on getting MariaDB to run on POWER recently, a bunch of which is based off my work on MySQL on POWER.
There’s obviously still some work in progress going on, but my initial results show performance within around 10% of MySQL, so with a bit of work we will hopefully see MariaDB reach performance parity.
One interesting find was the code to account for thread memory usage uses a single atomic variable: this does not scale and does end up showing up on profiles.
I’ll comment more on the code in a future post, but it looks like we will have MariaDB being functional on POWER in an upcoming release.
It was great to see the recent announcement of MySQL 5.7.5 over at the MySQL Server Team blog. I’m looking forward to throwing this release at some of the POWER8 systems we have for a couple of really good reasons: 1) Does it work better than previous MySQL 5.7 releases “out of the box” on POWER? 2) What do the scalability improvements in 5.7.5 mean for peak QPS on POWER (and can I set a new record?).
Looking through the list of changes, I’m (casually not) surprised as to the number of features and the amount of work that echoes what we were working on in Drizzle a few years ago.
A closer look at the source for 5.7.5 may also prove enlightening, I wonder how the MySQL team is coping with a lot of the code rot legacy and the absolutely atrocious internal APIs they inherited…
The following are not news:
(feel free to extend the list)
I want to emphasize this for those who have not spent time near trademarks: trademarks are trouble and another one of those things where no matter what, the lawyers always win. If you are starting a company or an open source project, you are going to have to spend a whole bunch of time with lawyers on trademarks or you are going to get properly, properly screwed.
MySQL AB always held the trademark for MySQL. There’s this strange thing with trademarks and free software, where while you can easily say “use and modify this code however you want” and retain copyright on it (for, say, selling your own version of it), this does not translate too well to trademarks as there’s a whole “if you don’t defend it, you lose it” thing.
The law, is, in effect, telling you that at some point you have to be an arsehole to not lose your trademark. (You can be various degrees of arsehole about it when you have to, and whenever you do, you should assume that people are acting in good faith and just have not spent the last 40,000 years of their life talking to trademark lawyers like you have).Basically, you get to spend time telling people that they have to rename their product from “MySQL Headbut” to “Headbut for MySQL” and that this is, in fact, a really important difference.
You also, at some point, get to spend a lot of time talking about when the modifications made by a Linux distribution to package your software constitute sufficient changes that it shouldn’t be using your trademark (basically so that you’re never stuck if some arse comes along, forks it, makes it awful and keeps using your name, to the detriment of your project and business).
If you’re wondering why Firefox isn’t called Firefox in Debian, you can read the Mozilla trademark policy and probably some giant thread on debian-legal I won’t point to.
Of course, there’s ‘ MySQL trademark policy and when I was at Percona, I spent some non-trivial amount of time attempting to ensure we had a trademark policy that would work from a legal angle, a corporate angle, and a get-our-software-into-linux-distros-happily angle.
So, back in 2010, Monty started talking about a draft MariaDB trademark policy (see also, Ubuntu trademark policy, WordPress trademark policy). If you are aiming to create a development community around an open source project, this is something you need to get right. There is a big difference between contributing to a corporate open source product and an open source project – both for individuals and corporations. If you are going to spend some of your spare time contributing to something, the motivation goes down when somebody else is going to directly profit off it (corporate project) versus a community of contributors and companies who will all profit off it (open source project). The most successful hybrid of these two is likely Ubuntu, and I am struggling to think of another (maybe Fedora?).
Linux is an open source project, RedHat Enterprise Linux is an open source product and in case it wasn’t obvious when OpenSolaris was no longer Open, OpenSolaris was an open source product (and some open source projects have sprung up around the code base, which is great to see!). When a corporation controls the destiny of the name and the entire source code and project infrastructure – it’s a product of that corporation, it’s not a community around a project.
From the start, it seemed that one of the purposes of MariaDB was to create a developer community around a database server that was compatible with MySQL, and eventually, to replace it. MySQL AB was not very good at having an external developer community, it was very much an open source product and not a an open source project (one of the downsides to hiring just about anyone who ever submitted a patch). Things struggled further at Sun and (I think) have actually gotten better for MySQL at Oracle – not perfect, I could pick holes in it all day if I wanted, but certainly better.
When we were doing Drizzle, we were really careful about making sure there was a development community. Ultimately, with Drizzle we made a different fatal error, and one that we knew had happened to another open source project and nearly killed it: all the key developers went to work for a single company. Looking back, this is easily my biggest professional regret and one day I’ll talk about it more.
Brian Aker observed (way back in 2010) that MariaDB was, essentially, just Monty Program. In 2013, I did my own analysis on the source tree of MariaDB 5.5.31 and MariaDB 10.0.3-ish to see if indeed there was a development community (tl;dr; there wasn’t, and I had the numbers to prove it).If you look back at the idea of the Open Database Alliance and the MariaDB Foundation, actually, I’m just going to quote Henrik here from his blog post about leaving MariaDB/Monty Program:
When I joined the company over a year ago I was immediately involved in drafting a project plan for the Open Database Alliance and its relation to MariaDB. We wanted to imitate the model of the Linux Foundation and Linux project, where the MariaDB project would be hosted by a non-profit organization where multiple vendors would collaborate and contribute. We wanted MariaDB to be a true community project, like most successful open source projects are – such as all other parts of the LAMP stack.
….
The reality today, confirmed to me during last week, is that:
Those in charge at Monty Program have decided to keep ownership of the MariaDB trademark, logo and mariadb.org domain, since this will make the company more valuable to investors and eventually to potential buyers.
Now, with Monty Program being sold to/merged into (I’m really not sure) SkySQL, it was SkySQL who had those things. So instead of having Monty Program being (at least in theory) one of the companies working on MariaDB and following the Hacker Business Model, you now have a single corporation with all the developers, all of the trademarks, that is, essentially a startup with VC looking to be valuable to potential buyers (whatever their motives).
Again, I’m going to just quote Henrik on the us-vs-them on community here:
Some may already have observed that the 5.2 release was not announced at all on mariadb.org, rather on the Monty Program blog. It is even intact with the “us vs them” attitude also MySQL AB had of its community, where the company is one entity and “outside community contributors” is another. This is repeated in other communication, such as the recent Recently in MariaDB newsletter.
This was, again, back in 2010.
More recently, Jeremy Cole, someone who has pumped a fair bit of personal and professional effort into MySQL and MariaDB over the past (many) years, asked what seemed to be a really simple question on the maria-discuss mailing list. Basically, “What’s going on with the MariaDB trademark? Isn’t this something that should be under the MariaDB foundation?”
The subsequent email thread was as confusing as ever and should be held up as a perfect example about what not to do. Some of us had by now, for years, smelt something fishy going on around the talk of a community project versus the reality. At the time (October 2013), Rasmus Johansson (VP of Engineering at SkySQL and Board Member of MariaDB foundation) said this:
The MariaDB Foundation and SkySQL are currently working on the trademark issue to come up with a solution on what rights to the trademark each entity should have. Expect to hear more about this in a fairly near future.
MariaDB has from its beginning been a very community friendly project and much of the success of MariaDB relies in that fact. SkySQL of course respects that.
(and at the same time, there were pages that were “Copyright MariaDB” which, as it was pointed out, was not an actual entity… so somebody just wasn’t paying attention). Also, just to make things even less clear about where SkySQL the corporation, Monty Program the corporation and the MariaDB Foundation all fit together, Mark Callaghan noticed this text up on mariadb.com:
The MariaDB Foundation also holds the trademark of the MariaDB server and owns mariadb.org. This ensures that the official MariaDB development tree<https://code.launchpad.net/maria> will always be open for the MariaDB developer community.
So…. there’s no actual clarity here. I can imagine attempting to get involved with MariaDB inside a corporation and spending literally weeks talking to a legal department – which thrills significantly less than standing in lines at security in an airport does.
So, if you started off as yay! MariaDB is going to be a developer community around an open source project that’s all about participation, you may have even gotten code into MariaDB at various times… and then started to notice a bit of a shift… there may have been some intent to make that happen, to correct what some saw as some of the failings of MySQL, but the reality has shown something different.
Most recently, SkySQL has renamed themselves to MariaDB. Good luck to anyone who isn’t directly involved with the legal processes around all this differentiating between MariaDB the project, MariaDB Foundation and MariaDB the company and who owns what. Urgh. This is, in no way, like the Linux Foundation and Linux.
Personally, I prefer to spend my personal time contributing to open source projects rather than products. I have spent the vast majority of my professional life closer to the corporate side of open source, some of which you could better describe as closer to the open source product end of the spectrum. I think it is completely and totally valid to produce an open source product. Making successful companies, products and a butt-ton of money from open source software is an absolutely awesome thing to do and I, personally, have benefited greatly from it.
MariaDB is a corporate open source product. It is no different to Oracle MySQL in that way. Oracle has been up front and honest about it the entire time MySQL has been part of Oracle, everybody knew where they stood (even if you sometimes didn’t like it). The whole MariaDB/Monty Program/SkySQL/MariaDB Foundation/Open Database Alliance/MariaDB Corporation thing has left me with a really bitter taste in my mouth – where the opportunity to create a foundation around a true community project with successful business based on it has been completely squandered and mismanaged.
I’d much rather deal with those who are honest and true about their intentions than those who aren’t.
My guess is that this factored heavily into Henrik’s decision to leave in 2010 and (more recently) Simon Phipps’s decision to leave in August of this year. These are two people who I both highly respect, never have enough time to hang out with and I would completely trust to do the right thing and be honest when running anything in relation to free and open source software.
Maybe WebScaleSQL will succeed here – it’s a community with a purpose and several corporate contributors. A branch rather than a fork may be the best way to do this (Percona is rather successful with their branch too).