Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror

Comment Re:Who is Planitir? (Score 3, Informative) 122

Palantir was founded by Steve Cohen, Joe Lonsdale, Alex Karp and Peter Thiel. Everyone of these people are some of the most over-educated people in the world. Alex Karp alone has a BA in philosophy, a JD from stanford, a PHD from Goethe University from Frankfurt.

Peter Thiel is a co-founder of Paypal where he made most of his early there before it was sold to Ebay.

Much of these people are staunch conservatives whom do not believe that a well rounded educated populace would benefit their interests.

Comment Re:Get Woke.. (Score 1) 204

The show isn't against interracial relationships.

It's talking about many black women against interracial relationships.

It's a real thing. On one hand, black women feel margainalized as their facial features are not considered beautiful by mainstream society. On the other hand, there are black men who actively light skinned people exclusively because darker skinned women are seen as less desirable, affluent and come from the wrong families.

Then you have folks who date whoever they find attractive and get caught in the cross fire.

But you should read the comments of the tweet, they are most illustrative.

Comment Re:How much smaller companies? (Score 1) 21

I think the weirdness of their names is confusing as well, but openstack has unified their userland interface to 'openstackutils' rather than having to contend with glance/newtron/cinder/etc etc's specific tools a few years back.

The incentive to openstack vs rolling everything on vms using proxmox is fundamentally some of the automation sugar that you can add on. A lot of the same basic stuff you can automate through aws, like security rules, networking, routing, firewall related stuff, can be orchestrated in a terraform manifest. You can create vm placement rules to make sure your vms are grouped as close together or as far apart as possible, e.g. try and put them all on the same virtualizer, or make sure only one vm lives on one virtualizer.

There are other services too, which if you are doing things like, k8s orchestration, standing up a quick hadoop2 enviromnent or even standing up backup and file sharing, there are services for all of that. You don't have to roll out a few vms and do your own configuration management on top to get the same thing.

Comment Re: How much smaller companies? (Score 1) 21

There is nothing theoretical about this. I've deployed OpenStack in production from 2012 to 2020. I've installed the whole thing by hand when I first had to learn it. I did the folsom to grizzly to havana to icehouse upgrade by hand. I've also stood up multiple clusters across multiple data centers in a matter of a few weeks. Ceph, IMHO, is a much harder beast to tame than Openstack when things go awry.

Much of keystone is pretty basic if you leave the things to defaults and only have to auth against active directory for one tree. Across multiple domains, it is best to use some config management, e.g. openstack-ansible to render things quickly.

I'm curious to know what you find so challenging about openstack-ansible when Juju+Charms and Tripleo are far more opaque.

Comment Re:How much smaller companies? (Score 2) 21

It's not that bad. The fundamental openstack apis are web services talking to unix userland. It is conceptually very simple. The more advanced services are web services that tied into openstack via openstack's own orchestration interface(s), i.e. webservices calling webservices. Much of the open source products underneath openstack runs on rabbitmq, apache, python, mysql/galera, memcache, haproxy and kvm, You do block storage through iscsi, ceph or whatever storage provider you want to employ, you do object storage through openstack's own object store, or ceph.

Openstack once upon a time aimed to be on feature parity with AWS. That is no longer possible, but terraform providers exist for openstack, as does deploying things through ansible/puppet/chef/etc etc.

Last but not least, you can orchestrate the whole thing through openstack-ansible. Change three yaml files, re-install a rack to base os through ipmi/pxe/cobbler and fire off openstack (ubuntu, rhel, centos, debian are supported), and kick off your deploy. You can know nothing and have a full cloud inside of a week. Most of the orchestrators, i.e. openstack-ansible, follow the deployment/hardening guide pretty religiously. You don't have to spend a lot of time locking stuff down if you control your management interfaces and lock that down at the network level.

Comment Modern skills (Score 5, Informative) 140

The demand for in depth knowledge of legacy services such as mail, dns have declined. Hardware troubleshooting is not as necessary as before.

A decent set of skills for systems administrators nowadays in the unix world include knowledge of configuration management, ci/cd, some mass orchestration framework like terraform and at least decent knowledge of one containerization frame work like docker and pretty decent know how of one cloud provider. I also would like development chops (which you have) as well as source control competency as well.

You'll want to know more about nginx instead of apache. You'll need to know systemd. If you did much of your scripting in bash and perl, bash is still in demand but python has largely eclipsed perl as the standard sysadmin glue. Standard appliances like F5 load balancers have been eclipsed to load balancers in the cloud, as it's just easier to scale.

Last but not least, since you've been out of the game for awhile and given your background, you might want to start working the MSP route and work your way up the chain again. It will be frustrating, but you'll see what is out there and you can work your way back into a modern skill set in a year or so.

Best of luck.

Comment Re:What's the replacement for FORTRAN? (Score 4, Informative) 205

Pleiades is just a slurm cluster. You can deploy new hardware, push your processes to be queued to the new hardware and link your run time to use cuda enabled libraries no problem.

Switching link time and link time hooks is pretty easy on Pleiades. It's the same way you would switch between python versions, different fortran run times (intel and gnu), different mpi runtimes (openmpi/sgi enabled stuff).

Source: worked on Advanced Vehicle Make and had to run cfd code on Pleiades.

Comment Re:Ursula LeGuin doesn't count? (Score 2) 252

Well, I think they are not the same thing. That's like going to a math conference and get disappointed that they don't hand out research awards for results found in mechanical engineering papers

Let's talk about two bodies of work that ought to be about the same type of people, but actually yield very different results. Let's compare James Salter's first two books, The Hunters and Cassada with The Expanse series.

James Salter's first two books draws from his experience as a late joiner to the American Army Aircorps, of which he graduated in 1945. The first book, Hunters, draws from his experience as a fighter pilot in the Korean War. It's about a guy who really wants to be successful as a fighter pilot. Throughout the book, while I the reader is constantly rooting for him, he is awash in self doubt. I don't think my efforts for rooting for him are in vain, but the character took awile for me to get to liking. Cassada on the other hand is about a guy who I imediately draw a liking to, but his charm does not extend to many of his cohorts within the book. I'm often surprised by what happens, how my heart strings get tugged with and it almost seems like disappointment hits both the characters and me in waves where I would least expect it.

But I read the Expanse series, and immediately the tropes jump out. As a novel where one co-author did the world building for an MMO game first, and then play tested it using D20 modern characters, the mechanics of this world seemed pre-ordained, the characters already seemed like they were a bit more sure of themselves and the people writing/playing them had a good idea on how they were going to react given a circumstance, as role player tend to have a habit of playing slightly different takes on the same theme most of the time. I open this book and come to feelings of simulteously being somewhere new, and having been there before, all at once. I remember a pair of adidas that I really liked playing basketball in when I was in my teens and in my late 20s it was re-released. The shoe felt similar, but better constructed. The Expanse feels no different.

Literary fiction toys with the literary fiction making process whilst in the guise of putting together a tale. The reader here does the work of deconstructing the process. The literary protaganist may be the center of the plot, but he/she/it isn't really driving it. Most of the time they aren't even all that likable in the beginning. I read literary fiction and examine how and why things are done while I anticipate what happens from page to page.

In commercial fiction, the main characters do most of the work for the reader. There's less playing around in style. Characters that drive the book are immediately likable. I don't really have to wonder about the why or the how. Most commercial fiction borrow important techniques from literary fiction from years past, and some sci fi writers read their literary fiction contemporaries (it's obvious that Neal Stephenson borrows from David Foster Wallace). So if I pay attention and keep up with my literary world stuff, the sci fi stuff will use dramatic techniques and plot devices that trickle down.

Comment It doesn't just happen on Apple (Score 2) 165

I've got an older GTX 760 running on an HP Z820. I run ubuntu on this thing and use nvidia-352 drivers. When I log out of gnome3 and log back in through lightdm, I see the same exact symptoms. I can see what was previously displayed on my framebuffer, including firefox and chromium windows.

Slashdot Top Deals

The bigger the theory the better.

Working...