Debian Sticking With Merged /usr Plan

[tuxd3v]

Kano ,
I sincerely appreciate your Humane and mature mediation, sincerely!
But there are several things that if you let them grow too much, that kind of people will jump on you..
And this hreindl its one of this cases,
First you ignore him, then he will continue insulting and trying to ashame you,
And that type of behaviour needs to be stopped in the beginning..
I am really sorry..

even in small companies these days you press a button and just restore the whole virtual machine which don't need to know about LUN, LVM and my full disaster recover plan in doubt needs one or more hypervisors mounting the isolated NFS storage hosting the backup appliance on a dedicated network or run the stuff from there and move it to the production strorage on-the-fly, desgined that way in 2010 or so,, we have 2019 now

but well, nothing to except from shitheads struggeling with systemd and Redhat in 2019

your environemnt sounds like it's grown too fast to be handeled as small and don't got the needed re-design to be be handeled as big as well, crap in other words

capable environments restore your crap from the outside because the operating system handeled by users has no access to backups by definitoon, they are done and restored from a layer below with a simple command or click by one of the few peoples which have access to that low layers

In Small bedroom companies, like yours, yes..
You are a MS Windows guy, with your "__mouse_clicks_to_restore__" ..

Can you tell me how to you restore a very complex system were its own OS was removed, and also all its luns data, were you don't even know what lun belogs to what mount point, or worst, even if some luns belongs to that machine..with a simple mouse click??

Please expand on that, so that we can laugh at it, even more..
Go ahead..

[tuxd3v]

*lol* not touched Windows for 12 years now at all

on a proper setup all the OS except the hpervisors (more than one!) are just virtual machines don't knowing anything about LUN
if you have no money and no knowledge then it's your problem

Now I am laughing at you..
So... so many trash talk, and what you know if kidding with virtual machines..

And you even have the guts to say that some companies don't have money?
Yours, are the one that don't have the money..because of that your small clients buy Vms there...its not the same but they are poor..

Because this machine in question is a Dedicated BareMetal Server, that receives luns on its HBAs from abroad..
Yes the Client Has the big money to run a Dedicated OS on it( its precisely the Opposite of what you are talking about.. )
And it runs a Custom OS config only for it..

when you don't have the money you kid with Vms,
But the performance..no way close to a REAL BARE METAL SERVER, running on the BASE OS...but this costs a lot of money...yes you need big clients for that..

So giving you another Opportunity..
How does you recover this Machine( Forget the Vmware kidding Solution... which is a shitty solution, by the way, but yes we also have tons of that sh**ty thing here.. unfortunately... )...?

on proper ssetups people like me write the scripts and admin interfaces to make such operations a single click in a webinterface calling prepared and tested scripts or take some money in their hands and setup EMC Avamar which perfectly works within VMware vSphere with no need for any Windows machine these days

We have all of the best products of IBM, EMC, and others,
Also we have dedicated worldwide Fiber Chanel Network, plus FCoE between datacenters, deduplicated backups, differential and all that stuff, and a lot more things than that,
Even logical instant images of hundreds of machines, yes almost real time for abroad Datacenter( But this Feature costs even more )..

unfortunately this one don't have real time imaging , so don't worry with that..

Instead of giving me bullsh**t again,
reply to the simple question that I made above.

[Kano]

Using bare metal servers with external iSCSI boot is certainly an interesting setup. I tested this 2012 with iPXE (and as alternative via pxelinux) without an iRMC for Linux (and even Win) with a very basic iSCSI Linux server. I guess most PXE boot setups for Linux boxes use NFS instead. Did you boot this way or did you still use local storage?

[tuxd3v]

i won't setup a machine like this because i am not an idiot and other like you i know that these days there is little to no measureable virtualization overhead but you get unfied backups, and hot-migration for free (next hardware generation, maintainance of the underlying bare metal...) and in case of huge virtual data disks mark them as independent so that you can recover only the operating system within seconds from Avamar while the data are backuped with a different solution

See your problem?
Does you see now the trash talk moron you are?

Because in this Comment,
I stated that I even needed to have a Debian 7 on my hands to save your RedHat Os..

Your Commnents:
"because you are an idiot?"

"so you shithead use a distribution with dist-uogrades every few years, can't handle it and think you act on my level doing major upgrades twice each year for a deacde with no downtime longer than a oridnary kernel update?"

Only Insults on that Thread, has if I were your Father or Brother or something like that, with no constructive content,
Only noise, lack of education, you act like a Lonely Wolf..

But you have lots of trash talk,
I only pushed this subject further to show, What/Why you are like that, and the amount of noise that was generated with that, after all, "without no brilliant Solution from you"( which I already expected, by the way.. )

But you are keen to label the Others as Idiots, and not professionals( ..in advance),
When you are, in fact,the real Idiot..

[tuxd3v]

Using bare metal servers with external iSCSI boot is certainly an interesting setup. I tested this 2012 with iPXE (and as alternative via pxelinux) without an iRMC for Linux (and even Win) with a very basic iSCSI Linux server. I guess most PXE boot setups for Linux boxes use NFS instead. Did you boot this way or did you still use local storage?

This machine was not setup by me,
In fact I didn't even knew that it existed before..
And usually, clients don't have administration rights,( but on this one, unfortunately was there a local account.. with root privileges )..
I simply used a Debian 7 iso
RedHat flavours doesn't worked there, or crashed, on boot, or doesn't had basic drivers for the HBAs( that were on that machine.. )
I can't recall now, but were a series of problems, that forced me( against the policies, to use RedHat only there ) and rely on the good old Debian 7, and then drop to a shell, and go by hand..

iSCSI could be a good option, for boot, disk-less systems
I think that we still have some machines( that don't require high bandwidth I/O ) using iscsi, but majority is FC or FCoE
Usually we don't have local storage, but that also depends on the setup.

[tuxd3v]

you won't get a soluton from me without a lot of money - it's that easy, i can only assure you in my setups this situation won't be possible because 100% virtualization and working backup/recovery plans

I understand that you come here to get some attention, and for sure you are trying to get mine..
But I am here for other purpose..

RedHat is a easy OS to use( in some perspectives..very limited in others ),
It has been improving a lot at least since RedHat 4
But has sh*tty things, that a power user that likes Debian, and have worked with Debian misses a lot..

Its not only me, other SysAdmins that I know, and followed Linux for some decades, says the same..
If I could describe it in a fast manner, it would be "hype"/"better supported by Corporate products"..

Accept it or not, its up to you..

[tuxd3v]

i don't give a fuck about attention at all

It doesn't seems so..

a power user can handle every OS to begin with

well, other sysadmins with a head full of shit like you, Weasel and your friends

the problem of idiots like you and "SysAdmins that I know, and followed Linux for some decades" is that you don't realize that the needs changed in that few decades, othes moved on, you don't - well, burry your head in the ass, nobody cares

Yes a power user handle every OS, even difficult ones like Debian or Slackware..
But I don't believe you are in this group..

You maybe can handle RedHat, Centos, and the likes, not technical ones..
You are more the kinda of guy that do "click_with_the _mouse", and that explains a lot why you like systemd, and think that Vmware, is the greatest thing ever..

Has per Weasel,
Its kinda of rude from you, to call, names on a person that was not even included in this conversation..

But What can I say, about a trash talk guy like you...so trash talk, but when I question you about a way of solving a problem you braked..
you showed everybody that you are indeed a Windows guy that started to kid with Linux, because you think it make you smarter..

You are years away from Weasel, in maturity, and in professionalism..

Also we use FC or FCoE,
Because we invested millions and millions on it, more in FC ofcourse..it was the first.. was a big mistake..
I think we only have 40Gb in legacy systems

majority Qlogic, but other also..

Infact this market is crumbling in a duopoly..
I found another provider:

https://www.chelsio.com

Any one, out there, has experience with chelsio HBAs?

[tuxd3v]

*lol* the guy not capable to hanlde Redhat systems talks about power users handling even difficult OS

Speak about yourself, I deal with them, for a long long time, but I prefer not to..
When you have Debian, why RedHat...I use it only because contracts requirements, not because its a great thing..

yeah, given that i wrote all the admin backends from scratch i am allowed to use them while i didn't see a GUI for systemd until now and it looks you are too stupid to handle symlinks and text config files :-)

I also wrote my Scripts, and Automation, no big deal on that...
And I do that for several Countries around the World, not like in your "bed room Company.."

Also, I advise any one starting( like you ), to start by that...
But I started doing that 2 decades ago..

And really, I don't want to do trash talk, this is only noise in the Forum..
Its a shame, and you are always pushing this thing on..
Which reflects your immature personality..

who cares, he is a similar fool like you with a lacking mindset for the IT at all

In first place I can only speak by myself,
But for what I read from him, he seems a very balanced person..

The same I cannot say about you...
with your emotional stability problems..only trash talk, in the moment to make the difference, you "crash and burn"!

i don't need to solve a problem that never would happen in a proper environment and no idea where all your Windows bullshit si coming from given that i didn't touch it since 2006 at all

That is bulsh**t... again

yeah, would you have used your brains before throw millions on it

Your Head is to small, to understand that at that time, we needed big amounts of speed, and no other technology provided the same, but still I was against it!!
The Decisions in Mega Companies are made at higher levels, we only give opinions about..
If you have already the Infrastructure in place its a great thing( low latency, and infrastructure in place already, downside big latency between ops, and big costs ), if you don't have it, forget FC, go FCoE or iSCSI with TCP offloading

40Gb legacy? laughable attempt to let your penis appear larger

Welcome to the world of MegaCompanies..

[Kano]

@tuxd3v

I did not fully get if you reinstall in case of problems or if you try to fix the system via a live system (and chroot) first.

@hreindl

A power user can certainly get used with any os. I even got used to write PowerShell scripts for Windows systems

Did you ever analyze a fully unknown system? That's what I did with Splashtop without the "normal" hardware...

[tuxd3v]

@tuxd3v
I did not fully get if you reinstall in case of problems or if you try to fix the system via a live system (and chroot) first.

You don't need a livecd, actually, the 1st CD of Debian7 is Just fine..to save the RedHat sh*ty OS..
I don't remember now about specifics, it was some 3 years ago..
But out of my head..

1) Restore OS
You simply drop to a shell( instead of install.. ), or we have the balls in place or not..
load the kernel modules for the HBAs
Configure correct IPs, and routing tables for frontend/backend/backup networks
find the Correct Luns, for each VG, for the OS, configure them,
Create the necessary VGs/LVs, with correct luns for OS ONLY.
mount LVs in a different File-system Place( ie: /restore )

configure Backup client, with includes and excludes rules,
Then carefully, restore to each LV/LUN(/restore/boot)

1.1) Before restart, disable all startup application daemons..and ofcourse the Vip address, on Recover Machine..
Also comment /restore/fstab, application mountpoints, let only OS mountpoints in place..( if not you will have a tremendous amount of time wasted on fsck on reboot, on application empty luns...yeah )

unmount temporary /restore filesystems

Reboot RedHat sh**T OS( this time it will have the correct HBAs drivers in place, so no problem..)

If ok,
2) Restore Application mountpoints..
Go sleep 2/3 hours because of fsck on all OS luns on startup..point 1) above..

Then go again and find the Luns for Application stuff..( around 60 LUNs ),
Uncomment /etc/fstab application mountpoints,

Verify ips/routing tables if they are in place..

Check mount-points/Luns if they are OK..verify for sure..
then mount Application File-systems

Then start restoring application Data to each LV mount point,
In the end, if you haven't sleep above keyboard( Because this types of restores are massive...even if you drop a line to Backup Team to give more priority withing your work...which I doesn't like, and his against the rules, also not fair for others.., but sometimes...you know... ),
reboot,

if ok,
3) Restore Databases/Application Layes
"go sleep for 2 days"..
The system, in the meantime on reboot, is very busy, doing a fsck, to any of the MEGA File-systems in the Server..

"After 2 days" it should be OK, log and verify if they are there..
Switch Roles for Slave on Database/Application Software....(see point 1.1) )
Start Databases, application software by hand, check logs, and if you are receiving commits from actual master

if ok,
4) Switch Roles of databases/Application Layers..

Stop databases/application layers, and enable application daemons..( disabled in 1.1) )
Also Comment again /etc/fstab, ( for application file-systems only....fsck on reboot, you have already done it..above )
reboot,
Then verify if all is in place, verify logs and if daemons started(and if they have started ok ..),
Report back to Application Team to make their own checks...

if ok,
5) Enabling the Restored Actual slave has Master

Stop accepting on actual Master Databases connection( in Redundancy Machine), check if all commits were applied
5.1) Disable Vip address in Redundancy Machine
Change /etc/hosts accordingly.. on actual master and Restores master

Apply all the commits from the old master( redundancy ) that are not in the restored machine..

If OK( you have not a split-brain or network partition problem.. ),
then,
5.1) Enable Vip address on Restored Machine
/etc/hosts accordingly
"go to sleep"

Maybe missing something here, for sure, has it was spitted out, in sequence,without much attention on it..

Replying to your question, directly,
No I didn't installed nothing, I just restored backups..

I don't reinstall Oses,
With so many applications and users for them, security rules, and so on,it was a very big machine,like many others here( But I never worked on that machine, I didn't knew its configs I didn't knew even it existed before.. )

Other fanboy arrogant RedHat SysAdmins here, refused to touch it..
Because RedHat doesn't recognised the HBAs...

if I would go reinstall this machine,
It would take me more than a 1-3 weeks to have all( Since it would be like a new project.. ), and even then lots of problems could happen..

Usually when I create a environment, with Production,Production_Redundancy,Quality, Development...
I write automation Scripts for them, which are around 10k lines for each project, some more, some less..

If I had that scripts at hand,
Maybe I could reinstall all, but then I will suffer with big amounts of time to load databases..
but all in all, using compression and so on, I could pass a database maybe in 8-9Hours( Our Network is fast... but it could be that at same time some of the zillion guys here could be doing some heavy stuff, and then the network could bottleneck a bit.. ) or so..

If very well coordinated( The Redundancy_Master will suffer with that, but could be feasible.. ),
Maybe it would be less time than Restore Backups...the problem.. no Documentation on that systems..
So Safest solution, go for Backups..

Like I sad before.. and I know what I am talking about,
A RedHat SysAdmin, messes with RedHat, only a bunch can jump in another systems!
It was needed a MAN, to jump there, and so I accepted the "Lord of the Rings" Challenge..

I knew in Advance, that Debian7 wouldn't fail me
Since I started in Debian.., I tested several distros, including the very Respectable SUSE,
But you know, "there are no love like the first one"

@Danielsan
Yeah Great times, great Rides there..

@hreindl
Actually, you are a clown,
And I exposed you above...
Too many trash talk from you, to many abuses calling names on people, you little bully..
When I asked you to go ahead and explain what you should do..."you burned and crashed"!!
I gave you 2 opportunities..

Too many trash talk, but you don't have what it takes...
you were speaking bad about Weasel, but you are not even half of him..sorry.