@openit Just say CentOS/RHEL in your profile. Then you're not lying - if a face to face interviewer ever pins you down on it you can honestly say that you favor CentOS over paying RH their pound of flesh.

@openit said in SAMIT: IBM Is Killing Off CentOS:

OpenBSD - Only two remote holes in the default install, in a heck of a long time!

You need to read that right. Key phrase being "default install". Not to take anything away from that, but mostly we use default install as a base platform for other stuff. And therein lies the rub. True, OpenBSD does a lot more due diligence than some when it comes to packages. But FreeBSD is no slouch either.

@openit said in SAMIT: IBM Is Killing Off CentOS:

@gotwf FreeBSD reminded me about OpenBSD. For what purpose or services you were using FreeBSD?

Not sure what it makes not to use OpenBSD or FreeBSD as CentOS and Ubuntu, even though they are known for solid.

Is that because continuous improvements in features, security and support?

As for OpenBSD vs FreeBSD, it is not an either or deal. I have been using OpenBSD since circa 2.5 (?) for more security sensitive applications, dns, firewalls and such. FreeBSD offers more versatility for general server use, more packages, and had the port system, wh/OpenBSD did not implement until much later. FBSD also has ZFS. For those interested in on the metal file systems, I guess I should mention DragonflyBSD and hammerfs. But who other than cloud providers are running anything on the metal server side these days?

@Pete-S said in SAMIT: IBM Is Killing Off CentOS:

FreeBSD is a very nice unix-like OS. More so than the linux kernel, GNU utils and various code that linux distros cobble together.

But it's not as versatile, for instance with hardware drivers. That's the drawback of having a smaller user base.

I'll concede the point if your main/only metric is versatility and your primary use case is desktop or mobile. Mine is not.

As for drivers otherwise, depends on what hardware. For years FBSD kicked arse on Linux server side. Especially early years, nics, raid controllers, etc. and pretty much all else server side - then Linux's niche as well. Early Linux tcp/ip was horrible comparatively, for example. Even as late as "Code Red". Linux got more traction, more developers, and they raided *BSD drivers for "inspiration" so they could call them new works and license GPL. Remember when the "real enemy" was M$ and NT. Once that victory became a fait accompli the penguinistas needed a new enemy: the BSD's and pretty much all else not licensed GPL.

Ah, the marketing wars and psyops fud. Linux had RH behind it. FreeBSD only was ever ROSS. Real Open Source Software. Curious, how many here used RH early-mid 90's. RPM's were like Slack, had to track down all depends manually. About then I discovered FreeBSD and what a godsend their package and ports system were. Clearly superior and light years ahead. Well.. we all know how that history panned out. Mindshare. Linux became the darling of the media. Meanwhile, Yahoo, then the world's largest web site quietly built out on FreeBSD.

Post Dot Com Bomb, FreeBSD really suffered. They'd embarked upon a major kernel refactor targeting SMP support and performance. The ensuing economic crash resulted in devs employed by Yahoo being laid off. Hence FreeBSD 5.x really sucked for too many years. Then 6.x and 7.x, although improved... still not there. So I bailed on FreeBSD myself somewhere during those years to OpenSolaris because I wanted all that ZFS goodness. Yeah, we all know how that story played out.

That's my take on the history of the 90's thru mid 2000's anyways. Ymmv, but hope you at least enjoyed the stroll....

But... times change. Fast forward to modern days FreeBSD performance is on par or better and OpenSolaris is no more. Yeah, Linux is going to provide better glossy pixel, interface with the latest gadgets, etc. Desktop/Multimedia experience.

Workstation wise, where I am using my box to admin other servers, and maybe also want to run some servers on my Workstation for ease of testing locally and not having to spend money spinning up vm's in the cloud, works great. Take it over Linux any day. Way more stable. Way more sane. And dev is not being driven primarily by for profit companies ready, willing and able to change up the rules at their whim.

So what is this big hardware versatility win? Mainly proprietary GPU's and maybe some sound drivers for higher end multimedia creator use cases. I'm presently sporting an AMD GPU in this box w/four display ports powering three monitors and bunches o' pixels. Works dandy. And didn't break the bank. That said, I am not a gamer. Don't care about creating multimedia. My workstation was mostly for work related tasks rather than play.

And then, once again, came the big bad wolf knockin' at the door: M$ turned over a new leaf, embraced FBSD and raided it for Azure. Now isn't that one a gas? I've not kept up. Am told that M$ gives back "some" to community but mostly holds back. And like a few here, I have good cause to be leery of big companies becoming too involved in FOSS. Cuz I got to ask y'all; "Is this IBM deal the first time you've been burned?". Thought not.

Heh, can you tell I am grouchy mood today? Procrastinating on other things. Apologies for the wall of text.

@Pete-S Set up by others prior to my tenure but to best of my recollection it was a single well provisioned server. After the Chef rewrite they switched to Chef paid version cuz PHB's wanted all that gui goodness.

Prior to that I'd done some smaller Salt deployments that I engineered. Heh, back when you had to write your own formulas. I really liked Salt at the time. Also favored Python over Ruby. Just cuz. Neither would be a deal breaker for me but they definitely were for some dev shops so you just gotta roll with it.

Had another gig with a Chef deployment comparable to one mentioned above. This would have been prior to the rewrite/refactor. Chef definitely required more provisioning. While I liked Salt better you just get used to whatever and deal with it cuz those decisions were above my pay grade and more oft than not based on "buzz" than actual technical merit and/or evaluations. Crazy way to do business, eh?

Ansible is the one I never actually used beyond tinkering. I favor ROSS, or as close as I can come, so I'd be taking a long hard look at Ansible today. YAML is also preferable to Chef's DSL so that's another "win" for Ansible from my perspective. If you trust Salt security side, probably be my second choice. Caveat being that I've not used this stuff seriously in recent memory, so take all this w/a grain of salt. Heh, oohhh, two unintentional Salt Stack puns in a single sentence.

@scottalanmiller Lots of newness in 16.x. Been kicking it's tires a bit myself as of late. But... ya' know... 1.17 is almost here!

Congrats & Happy Anniversary!

@scottalanmiller Iirc, there is something known borked w/emojis on latest 1.16.x Fixed in soon to be released 1.17? Think there is a bug filed at github about it if you're curious.

Groovin' on some Cosmic Gate.

@Pete-S Nope. Don't sling code. That said, worked with lots of devs providing ops support. The extremely bright genius types that stand out and really had my respect, each technical team leads, were using one of:

• VS Code
• IntelliJ EDEA

And notably, they'd seem to swap back and forth from time to time. At least long enough to trial new features of new releases. Big Company was footing the tooling bill so cost was not a consideration for them. Ymmv.

@eddiejennings Could oft as well have both. .profile is more generic while, duh, .bash_profile is bash. .profile could for e.g. be used to read .kshrc. So groove on it...

@pete-s +1 on the patch panels and the catV back to main switch for low cost budget solution.

Rockin' some Etta

Etta James - Somebody To Love (Let's Roll)

@scottalanmiller said in Misc go-to FOSS options:

Percona is a DB support vendor, not a database maker. They provide packages and support for MySQL, MongoDB, PostgreSQL, etc. But it's just the other databsaes, with Percona's support.

Nope. Their xtradb and clustering stuff was (cuz I have not used in a while since my got to is PostgreSQL RDBMS side) much nicer experience than anything I had used MySQL based. Maybe this has been pulled into MySQL and/or MairaDB by now? Once again, I try to stay away from that entire ecosystem modern times. I like BSD licenses.

So maybe at least take a gander before dismissing: https://www.percona.com/software/mysql-database/percona-xtradb-cluster

Maybe not too many folks here need to concern themselves with clustering? Just throwing it out there in effort to be helpful. No skin off of my nose either way.

@scottalanmiller said in Misc go-to FOSS options:

We moved to OpenLiteSpeed. It's noticeable how much faster it is.

Interesting. I used Lightspeed bitd, before they had yet even a logo. I guess it went commercial? Stopped using it in favor of nginx when they hit a bunch of security issues. This was long ago. Have not looked at it since.

Up and coming new kid on the block that I am looking real hard at web server side:

• Caddy

But I have not yet deployed it so omitted from my initial reply.

There are many reasons I prefer to avoid Linux these days. I don't have time, nor interest, in an exhaustive list, since I am pretty sure NIH will prevail regardless. Perhaps one of the most attractive aspects of Illumos based stuff are the communities around them. Very helpful and welcoming. Takes me back to the good 'ol days in the 90's before FOSS had become so corrupted by special interests.

The hostility of many (most?) Linsux groups is one of the major things that have morphed Linux into Linsux. It has to do with size and mass adoption leading to developer/helper fatigue, one upmanship gamesmanship, etc. I see FreeBSD slipping and sliding down this slope as well. Sad days. OpenBSD has always been a bit "edgy" - you'd better have done your homework before posting, always been welcoming to newbies, but also a low tolerance for entitlement attitudes expecting "hand holding" w/o putting in any efforts/minimal on their end.

As for Debian... way too political and these days pretty much a serf of Canonical. Besides that, the "Debian Way" does too much tweaking for their own justifications. Makes troubleshooting more complex, relatively speaking, to something staying "more true" to upstream, e.g. Archlinux.

Been bit one too many times in the past by 'buntu though. Brain dead stupid stuff that caused lots of grief. It may well be great. For now. But I cannot trust is any longer.

As for ZFS... I am well familiar with ZOL having been an early adopter long before Unbuntu embraced and bundled. Heh, BTRFS was the long hoped for and awaited, waiting, more waiting Linux darling of the day. And boy, ho boy, did I get flamed on #archlinux whenever I would bring it up. Fast forward a bit and some of those same folks now asking me for "recipes". So I posted up a blog post on it, which I could link here for shits and giggles but, heh, that stuff is so old and dated that I have not yet even bothered to convert to https....

My two bits worth. Your mileage will likely vary. I endeavor to be objective but life experience will still leave us with prejudices. Sometimes that is called "learning". Others closed mindedness.

@bbigford said in Misc go-to FOSS options:

I've been wondering about others' preferences on a few things. The landscape of operating systems and databases has changed a bit in the last few years. Not curious about Windows or MSSQL, nothing new there.

Server OS: I've bounched back and forth with CentOS before Stream (the split between 6 and 7 was weird), Ubuntu Server (seems to get a lot of hate, no idea why), Fedora Server (also seems to get some hate, not sure why), RHEL (only when the customer absolutely requires the support and can't convince them otherwise), Debian (not used a ton, not sure why, pretty barebones)

I favor native ZFS based systems server side:

• FreeBSD because I am sick of the Linux corporatocracy pushing inferior technical solutions, b.s. internal power play politics, yes (wo)men shills, etc.

Otherwise something IllumOS based:

• OmniOS

• SmartOS

NoSQL DB: MongoDB went through a really shady legal bit when they were doing their as-a-service initially, which basically spelled out they own your IP if you use their DB with your app, haven't checked back to see if that got cleared up. On-prem I've used CassandraDB and MongoDB mainly, and started looking into ScyllaDB more recently.

No preference here, not a dev, so whatever is called for. But, yeah, two thumbs down on Mongo shady licensing change 'chit.

TSQL: Defaulted to MySQL until some devs spun off concerned with the Oracle acquisition and started defaulting to MariaDB

• MySQL replacement: Percona

• Else otherwise PostgreSQL. I favor ROSS: Real Open Source Software.

What are some of your go-tos these days? Why?

Others that scratch my various itches:

Web: Nginx

IMAP: Dovecot

SMTP: Postfix

But surely all of this is old news? Or maybe start a popularity poll?

Have fun!

@pete-s +1 on the patch panels and the catV back to main switch for low cost budget solution.

@stacksofplates Ah yes, Ergodox. Massdrop? Getting stuff off pinkies and onto thumbs is a big win. Alas, I don't have such a rig myself. Fairly pricey, iirc, no?

@pete-s said in Ergonomic Keyboard:

I don't believe there is any data that supports that assumption.

Plenty. It is called empirical evidence collected by me over decades of keyboarding. This is not rocket science. Get a keyboard. Type on it for a few months. Get a different one. Compare and contrast. Wash, rinse repeat. If you're going to spend anywhere near as many hours per day behind a computer as I over the years, it behooves one to make an investment in their workstation. And I am not talking about the latest and greatest Ryzen's. But you've got to do this consciously. Try to find the sweet spot in the switch. What were the designers thinking? For me, the arrival shock of bottoming out seems to aggravate my Dupuytrens. Using Cherry MX mechanical switches helps me be more aware of my keyboarding technique and that greater awareness seems to translate into the real world as greater periods between surgeries.

Your mileage may vary. I was responding to a specific query from the OP. Then the rest of y'all macho types decided to dog pile on. But trouble is that yer' wrong. And right. Because the key is not whether something is branded as ergo or not but rather underlying causes. And if some gizmo like a better chair, monitor riser, foot riser, vertical mouse, mech keyboard provides even perceived benefits by the users? Hmm...who's to say whether that is effective or not, eh? Geeze, Louise! IT geeks make this huge investment in "professional development" every day just trying to keep minimally abreast. Make some investment in your physical well being and set up a decent office space. It's not that hard. Nor costly.

Be all that as it may.... I would posit that the biggest win for a mech keyboard is the typing experience itself. They just feel nicer. And I make lots less errors. In no small part, no doubt, because I am used to that sweet activation spot on these cherry browns. Scissor switches common on laptops are okay. Better than rubber domes. But still need to bottom out to activate. Cherry MX, Buckling Spring, Scissor, Rubber Dome, Topre Hybrid. Use whatever feels best. Who cares if the rig is $20 or $200? It does not matter over the decade you'll be using it. Write it off in three.

@scottalanmiller Yes. This I know. But this isn't bare metal but rather SmartOS hypervisor in a Triton datacenter. So I presume all is atop ZFS. I never delved deeper so honestly do not know.

@pete-s Interesting that DAS is so anti ergo when they benefit from a faction of the market looking for Cherry MX switch'ed boards. Which, off course, are, and have been, DAS's bread, butter, and jam.

So we have all these feelings being summarized. Blah, bleah. Feeling Madame? I know not feelings. Have you any data? If what you have works great for you then that is great for you. That does not mean that your singular experience extends to the universal.

As mentioned above, everything is connected to everything else. So after your done making excuses and have your workspace set up at least half way decently, worked on minimizing PEBKAC related, then these non commodity keyboards have their utilities.

I have used, but do not use as daily driver, tented Goldtouch keyboards. Prevention? Meh! This is for somebody who's gone a bit past that and lookin' to mitigate to some extent. And therein I've had favorable reports from carpel tunnel sufferers.

Myself, am a skinny guy so straight keyboard works for me. Ten keyless and Otaku, if you please. The switch is the distinguishing factor. Of course, becoming aware of and adapting your key strokes themselves to mechanical switches requires self awareness. And some practice. Practice to unlearn the "bottoming out" that is fact of life for a rubber dome switch to fire. Which of course leads to "pounding" the keyboard. At least by some. Be especially aware during times of stress.

So switch comparatives like to cite activation forces. And maybe switch travel as well if you are lucky. Ergo marketeers will push those low activation key switch as big 'feature'. And maybe they are for some.

But for me.. it is more about "arrival shock". And a with a good MX switch and some practice you can pretty much all but eliminate this one since the switch provides tactile feedback on activation and does not require bottoming out to fire. Seems to be the sweet spot for me and my particular issues.

Edit: Fortunately, the engineering of rubber dome switches themselves provides a cushion for arrival shock. But rubber dome switches are "mushy" comparatively. Also, if you switch to an MX based board do be wary of bottoming out cuz that rubber cushion is now absent.

@pattonb Well... I have been doing some testing recently.... I have a NodeBB instance that I need to migrate off CentOS-7. Reputedly, although mongodb dropped support for else but linsux, the freebsd mongodb port runs great. Cool. I have been watching and waiting with bated breath to test this out. Unfortunately, most all hosts hosting fbsd do so via kvm, wh/puts fbsd under the bus from the get go. Contrarily, fbsd's bhyve stuff is reportedly superior ... so.... now that I had access to latest and greatest fbsd-12-patchset I decided to put through the paces.

All was same, on minimally provisioned servers. Insanely minimal. But it is a kind of perverse stress test. VM's were courtesy MNX.io, g1.nano.

Conclusions: Shooting, as always, from the subjective hip:

1. MongoDB was consistently a tad bit faster on Debian 10.

2. MongoDB succeeded in crashing itself during one of the FreeBSD perf tests. I tried a bit of tuning, wh/helped a bit... but not enough and still crashed.

3. MongoDB dropped portability and has been coded Linuse only fore a while now. So not to be unexpected that would run best on Linux. So I'mma gonna' use Debian 10. Cuz I'mma not a fan of 'buntu, okay?

@scottalanmiller Hm... I'd thought of posting that up here but figured it'd be old news for you lassi it hounds... Good news for the battle for ROSS.