Andrew Savchenko <bircoph <at> gentoo.org> writes:
> On Sat, 17 Jan 2015 21:04:44 -0500 Rich Freeman wrote: > > Speak for yourself. :) I did comment on my thoughts in this area in > > Donnie's thread. Gentoo (IMHO) tends not to be the best distro for > > doing anything in particular. I find that its best feature is that it > > is reasonably good at doing just about anything - it is a > > jack-of-all-trades. I think the fundamental flaw with systemd is the fact that the duality of support for systemd and other init solutions is so quickly abondoned. If they were allowed (encouraged) to run side by side for a few years, let folks decide then; as it is a major abandonment of principal, imho. Lot of folks in the embedded linux world, are scratching their heads at systemd; the conclusion from most of what I read is "no thanks anyway". > I can't agree with you here, though your position have a rationale. > I see Gentoo as a Universal Constructor (UC) which may be used to > whatever specific needs Linux can be used at all. +1, and more. It encourages folks to "dream". As a mostly hardware base guy, gentoo has inspired many things in the embedded realm, minimized systems and the future of 64bit embedded systems. Gentoo is at the heart of the embedded-distro collision and will manifest as 64 bit arm solutions reach street pricing. > In general UC pros is ability to create setup suitable for every > specific need, but cons is maintenance cost to create and update > such setup. Also creating and maintaining UC-powered setups rises > general professional level of system architect or amdin doing the > job. I post a thread a while back, from a current forum where I was surprised at just how many folks are maintaining hundreds if not thousands of (gentoo systems) as a superior solution for their needs. The one thing I took from that posting from literally dozens of folks is those that go down the massive gentoo deployment path are (1) very compentant to say the least (2) rarely contributed back to the gentoo distro. The fact that they do not contribute back, is not due to selfishness, but the "cost barrier to entry" for them to contributed back. Maybe a second gentoo wiki is needed for folks to just "put their ideas and files and scant directions up, without such a rigorous formality to contribute back? This, for me is the saddest part of the entire gentoo echo_system, ymmv. > So everything comes to how much user needs deviate from what > already existing binary distributions provide. If user needs are > perfectly satisfied with some binary distro, using Gentoo will only > raise maintenance costs. But if users demands something hardly > achievable with other (binary) distributions, then this is a good > place for Gentoo. I think you drasitcally over_estimate the number of those happy linux distro users. I think if there was an easy way to perform a few typical gentoo installs (workstation, mail-server, web server, dns server, hardended*) then folks would migrate heavily towards gentoo. I think Alan and his ansible_installs has the mindset for an experimentail gentoo-ansible install engine; not for all possible tweaked installs but certainly some of more (amd64) common installs. The questions is will someone who get anisble_based_gentoo_install working be afforded an "encouraging mechanism" to set up such a limited experimental installation semantic for new gentoo installs? > From my own experience I can point three directions where Gentoo > was and is reasonably the best choise for our needs (mine or my > colleagues): > 1) HPC. When it comes to scalable tasks and large amount of > hardware, even small performance gain results into huge saving of > costs. On our first cluster we replaced CentOS by carefully > tuned Gentoo and performance gain was about 30-50% depending on > scientific application (please note I'm talking about real > applications and not about synthetic tests like linpack). With > hardware costs about million of dollars, 30% performance gain > results in a great saving. Price for that was much longer time for > initial setup (many weeks instead of many days), but it was > still less then time required to setup hardware itself and all > auxiliary engineering systems. Donny alluded to this recently too in his planet gentoo posting. I am just saddened that HPC/Distributed herd/project has become so inactive. Yes, I am working in this area, but it has become vast with codes to test, split among systemd and other init components and the landscape is very noisy from the vendors offerings of "half_baked" open source musings and offerings. For me alone, it is a bit daunting, to say the least. > An interesting observation here is that average software update > cost of Gentoo is smaller that one of RH-based systems we used > before. While it is easier to update RH-based solution within the > same branch, then Gentoo setup, it is a complete nightmare to > upgrade from one branch to another, e.g. from RHEL4 to RHEL5. I've > gone through such update in the past an it is much worse than remove > everything and install from scratch, including all user > applications. As for Gentoo, all updates are equal: they bring some > build failures, runtime issues and compatibility problems, but to > a limited extent, which is handleable easy enough by prepared team. Continuous Integration (CI) is one of the keen areas of development for one of gentoo's derivative distros (Zentoo). Another area where Gentoo seems to be out front of the other linux distros, imho. > 2) High security servers. We have some systems dedicated to a very > specific needs where security demands are extreme. Hardened Gentoo > is the best solution here, since we can strip down such system close > to an absolutely possible minimum and protect that minimum by all > means (hardened toolchain and flags, PaX, SELinux and so on). Of > course, on top of then containers may be use to isolate different > daemons and so on... +1 > 3) Individual interested in getting every bit of performance > possible from own hardware. Frankly this was the reason why I > switched to Gentoo from RH about 8 years ago. I just tired to > rebuild each time a significant part of packages with custom flags > and configure options. Gentoo is much better suited for this task. > And as a result 13 years old hardware is still usable to watch 720p > and most of 1080p videos (without GPU hardware decoding). A > byproduct of such interest is a deep understanding of system > internals, which is a great result on its own. Yes, additionally: This is identical in embedded systems; except is it mission critical. Getting a lower cost microP to run a collection of codes (for a product or service) often means the difference in profitability and failure for an embedded project. Gentoo accels in this mode of deployment. Yet our embedded public presence seems to be very quite the past few years. The shear number of cool, new gentoo derivative distros does not only suggest that it is strong, but that it is a platform of supremacy for innovation; and that my friend is the very lifeblood of Unix->BSD->Linux imho. Gentoo does have a high cost of expertise to become useful to a *nix admin/hack/coder. Maybe with the new wiki and some of the newer developments, we can offer up rapid install solutions for many folks in the linux world. That is my hope. I think Alan is a pioneer for suggesting ansible cookbooks etc for gentoo specific installation needs. I hope he posts his works for us all to follow? Interestingly, Bircoph has solve many of the problems that seem to be in my path of discovery. Very, very cool and encouraging. I think Sven is giving gentoo new life, by his heroic efforts to create excellent documentation in everything he touches! > Best regards, > Andrew Savchenko peace, James