Linux Pioneer Munich Confirms Switch to Windows 10

[Megalith]

After more than a decade of running Linux-based PCs, Munich's city council has decided to switch about 29,000 PCs to Windows 10. Back in 2003, the council decided to switch to a Linux-based desktop, which came to be known as LiMux, and other open-source software, despite heavy lobbying by Microsoft. But now Munich will begin rolling out a Windows 10 client from 2020 at a cost of about 50m.

Past surveys have found only a minority of staff wanted to return to Windows and Microsoft Office. However, there have been vocal critics of IT, with the human resources department saying productivity had "decreased notably" due to crashes and printing errors since moving to open-source software. One council insider with knowledge of the LiMux described the decision to TechRepublic as "tremendously disappointing," while another said it was "a disaster in terms of costs.”

 

[Frobozz]

The council also backed a 6,000-seat trial of Microsoft Office 2016, which will be run on virtual machines.

Oh! They even bit on VDI! Sales execs are going to have a hard time deciding what yacht to get.

 

[dandirk]

sounds very familiar... I work in a medium sized local municipality only slightly larger.

All complaints sounds familiar and we are a MS shop...

Part of their problem is customization and another part is lack of leadership most likely. If HR is complaining about printing and that is having an affect on organizational strategy, you have problems imo.

On the flipside, you can't ignore the challenges of trying to find solutions/software when you basically take the most supported/used vendor off the table (or require least common denominator of support).

Customization is an issue as well... If they really have 12k custom templates and macros for office documents they are doing something terribly wrong to begin with. We are trying to remove customization at every opportunity. Mobility and the speed of technology these days practically requires more agility and customization hinders this. If needed at least put it in the most compatible format possible, HTML etc.

 

[Jim Kim]

"crashes and printing errors"
More like "user and hardware" errors.
But don't worry, those c&p errors will soon be replaced with malware, adware and ransomeware errors along with the increased cost's for running redmond. joy

 

[quasar56]

Money talks!

 

[MV75]

"crashes and printing errors"
More like "user and hardware" errors.
But don't worry, those c&p errors will soon be replaced with malware, adware and ransomeware errors along with the increased cost's for running redmond. joy

They'll wish they had that again when the entire network is down due to a weekly botched forced update.

 

[Shmee]

They'll wish they had that again when the entire network is down due to a weekly botched forced update.

As an IT worker in a large Windows shop, that has never happened for us.

 

[M76]

As an IT worker in a large Windows shop, that has never happened for us.

Sure, because not even a windows shop is as stupid to leave the actual network infrastructure to w10.

In other news one of my w10 installs just fried today when it tried to install a gpu driver trough windows update, even though driver updates were disabled in group policies. But it still tried to install the microsoft branded driver for the new gpu I put in the computer. Breaking the entire driver framework, to the point that I couldn't install any driver for any GPU thereafter.

 

[jwcalla]

I feel like this was posted here last week.

 

[DPI]

Money talks!

More like FREE talks. MS got on their hands and knees and said "we'll give you all the licensing and support free, please please just announce you 'chose' to go back to amazing Windows 10 with Edge and Cortando, plus Office365 and the Power of the Azure Cloud ™"

This is about as authentic as when MS throws a bunch of Surface tablets at the NFL or an airline for the propaganda value. Never works. But Windows installbase is in decline so can't blame them for trying.

 

[Hakaba]

They'll wish they had that again when the entire network is down due to a weekly botched forced update.

That Sysadmin should be fired for not properly testing patches before mass deployment to a production network.

And before anyone else goes off on how shit Windows is and all of your Linux boxes have run for 30+ years with no downtime. Stuff breaks, either drop a new PC, reimage through MDTS/WDS/SSCM, and move on.

 

[dandirk]

More like FREE talks. MS got on their hands and knees and said "we'll give you all the licensing and support free, please please just announce you 'chose' to go back to amazing Windows 10 with Edge and Cortando, plus Office365 and the Power of the Azure Cloud ™"

This is about as authentic as when MS throws a bunch of Surface tablets at the NFL or an airline for the propaganda value. Never works. But Windows installbase is in decline so can't blame them for trying.

Devices used to run a municipality and order taking ipad/iphones on airplanes are 2 different things.

FYI Advertising and marketing works, that is basically what you mention are. Pay x dollars for an opportunity for getting customers... No clue how often it works or not, doubtful it is "never".

The Delta airline example is interesting... I personally believe Apple is successful in these areas due to its simplicity and walled garden approach. It just works, worse case wipe and reload... A lot of Win10 is MS trying to move to something similar and what a lot of people in this very forum complain about.

 

[hescominsoon]

After more than a decade of running Linux-based PCs, Munich's city council has decided to switch about 29,000 PCs to Windows 10. Back in 2003, the council decided to switch to a Linux-based desktop, which came to be known as LiMux, and other open-source software, despite heavy lobbying by Microsoft. But now Munich will begin rolling out a Windows 10 client from 2020 at a cost of about 50m.

Past surveys have found only a minority of staff wanted to return to Windows and Microsoft Office. However, there have been vocal critics of IT, with the human resources department saying productivity had "decreased notably" due to crashes and printing errors since moving to open-source software. One council insider with knowledge of the LiMux described the decision to TechRepublic as "tremendously disappointing," while another said it was "a disaster in terms of costs.”

part of the problem is munich tried to build their own Linux distro AND implement it....bad move. While it is too late they should have gone with either Redhat or ubuntu...then they would have had true suport from implementation to user support...something they did not have building their own distro.

 

[MV75]

That Sysadmin should be fired for not properly testing patches before mass deployment to a production network.

And before anyone else goes off on how shit Windows is and all of your Linux boxes have run for 30+ years with no downtime. Stuff breaks, either drop a new PC, reimage through MDTS/WDS/SSCM, and move on.

Sysadmin and testing? Choosing to block certain patches out of a cumulative update indefinitely? How quaint. This is windows 10, not 7. Oh yes, they all will be moved to a surface as well. That breaks all the time. They'll have great up time, progress baby. Just move on and forget that there was work to be done, nothing important, just running a city. Get over it because our generations' embrace of the sub standard is never wrong and the vendor helps our heads not hurt by having to think.

 

[BloodyIron]

I've been kinda keeping up with this, and the thing I don't get is... Why did they not explore a hybrid-environment???

You can literally have parts where it's Windows and parts where it's Linux. Hell, you can serve the Windows-Only apps through Windows Server, or Citrix, TO LINUX USERS. Know how I know? IT'S MY JOB.

This is nothing less than Microsoft buying out the contract. Microsoft is opening a HQ in Munich, to say that is irrelevant is ignorant and delusion.

 

[BloodyIron]

Why didn't they just use CentOS or Ubuntu and adjust the settings to their environment? Maintaining a distro is A LOT OF WORK.

part of the problem is munich tried to build their own Linux distro AND implement it....bad move. While it is too late they should have gone with either Redhat or ubuntu...then they would have had true suport from implementation to user support...something they did not have building their own distro.

 

[gamerk2]

Why didn't they just use CentOS or Ubuntu and adjust the settings to their environment? Maintaining a distro is A LOT OF WORK.

Requirements. CentOS or Ubuntu probably didn't meet their internal requirements.

 

[velusip]

part of the problem is munich tried to build their own Linux distro AND implement it....bad move. While it is too late they should have gone with either Redhat or ubuntu...then they would have had true suport from implementation to user support...something they did not have building their own distro.

LiMux is just Debian, heavily configured. Regardless, the more I look at the history of LiMux, the more it seems like an elaborate cash grab. And why not target a government? Bottomless coffers, very long term (migration and integration), cushy job, federal client expectations (lol), multimillion dollar price tag, and a client willing to drop 13million upfront... yeah, when a state gets played, it doesn't figure it out for a long time.

 

[Hakaba]

Sysadmin and testing? Choosing to block certain patches out of a cumulative update indefinitely? How quaint. This is windows 10, not 7. Oh yes, they all will be moved to a surface as well. That breaks all the time. They'll have great up time, progress baby. Just move on and forget that there was work to be done, nothing important, just running a city. Get over it because our generations' embrace of the sub standard is never wrong and the vendor helps our heads not hurt by having to think.

Reading problems? Where did I say to block the patch indefinitely? I didn't. But if it has to be done then so bet it. There should be several other layers of security that lay between the user, the servers, and the outside world.

What I am pointing out that anytime I've seen a patch, program update, or the likes break a program/system installation its almost a day or two later (maybe a week) that update is fixed. That proper testing on a (test) network may have identified this, resulting in little to no downtime for your network, loss of data, money etc... The whole 1 patch will bring down an entire windows network could happen if when things are not done properly, same would apply to any other OS. Nothing is immune to poor practices.

 

[BulletDust]

Requirements. CentOS or Ubuntu probably didn't meet their internal requirements.

Considering how modular Linux is and the level of customization available to the OS, I find that highly unlikely.

 

[BulletDust]

That Sysadmin should be fired for not properly testing patches before mass deployment to a production network.

And before anyone else goes off on how shit Windows is and all of your Linux boxes have run for 30+ years with no downtime. Stuff breaks, either drop a new PC, reimage through MDTS/WDS/SSCM, and move on.

Spoken like a true corporate Windows sysadmin.

 

[B00nie]

As thought this was a political decision i.e. Microsoft paid someone off: https://www.techrepublic.com/articl...son-to-return-to-windows-says-citys-it-chief/

 

[dgingeri]

Sysadmin and testing? Choosing to block certain patches out of a cumulative update indefinitely? How quaint. This is windows 10, not 7. Oh yes, they all will be moved to a surface as well. That breaks all the time. They'll have great up time, progress baby. Just move on and forget that there was work to be done, nothing important, just running a city. Get over it because our generations' embrace of the sub standard is never wrong and the vendor helps our heads not hurt by having to think.

After being in IT for over 20 years, I can say you completely missed the point. The process is this:
1. Test the updates on a small subset of systems, preferably non-production systems
2. If one update has issues, do not approve that update for production, rather wait until either Microsoft fixes the update or, if a third party program is the issue, get the fix for the app, even if this means blocking a cumulative update. This does not usually take too long. It is definitely never "indefinite".
3. Proceed with updates

I know this process directly. I work with 60+ servers where I have to track their updates and got through this, weekly. As a matter of fact, I get to do this again in a few hours. We have never had an update "take down the whole network." We have had issues with our internal apps, but we have our own developer group, and they get things fixed pretty fast. That's why we have a QA group.

We have, however, had a malfunction in group policy that ended up forcing several desktop systems to reboot in the middle of production use. We had to refund some money to customers because they didn't some of our services in a timely manner. I doubt a city government will have issues with that. We did an analysis about the cause, and someone was found at fault but it was an easy to miss mistake. Rather than edit an existing policy, he created a new one to make some changes to existing update schedules, and did not grant the right permissions when implementing it in production, so the affected systems went back to the default update scheme. The desktop systems that had not been left on overnight to update forced an update on the users the next day, and the reboots took up to 15 minutes. (We do tell users to leave their machines on overnight to do updates, but many don't listen.)

So, in essence, doing things right, we never encounter what you talk about here. The trick is not being incompetent.

 

[tetris42]

After being in IT for over 20 years, I can say you completely missed the point. The process is this:
1. Test the updates on a small subset of systems, preferably non-production systems
2. If one update has issues, do not approve that update for production, rather wait until either Microsoft fixes the update or, if a third party program is the issue, get the fix for the app, even if this means blocking a cumulative update. This does not usually take too long. It is definitely never "indefinite".
3. Proceed with updates

I know this process directly. I work with 60+ servers where I have to track their updates and got through this, weekly. As a matter of fact, I get to do this again in a few hours. We have never had an update "take down the whole network." We have had issues with our internal apps, but we have our own developer group, and they get things fixed pretty fast. That's why we have a QA group.

We have, however, had a malfunction in group policy that ended up forcing several desktop systems to reboot in the middle of production use. We had to refund some money to customers because they didn't some of our services in a timely manner. I doubt a city government will have issues with that. We did an analysis about the cause, and someone was found at fault but it was an easy to miss mistake. Rather than edit an existing policy, he created a new one to make some changes to existing update schedules, and did not grant the right permissions when implementing it in production, so the affected systems went back to the default update scheme. The desktop systems that had not been left on overnight to update forced an update on the users the next day, and the reboots took up to 15 minutes. (We do tell users to leave their machines on overnight to do updates, but many don't listen.)

So, in essence, doing things right, we never encounter what you talk about here. The trick is not being incompetent.

So are you saying you don't have a plan if an update causes a problem that simply doesn't get fixed? On Windows 7, you simply skip that particular update. On Windows 10, you cannot update again if that happens. I'm not saying you've ever had that problem (I have, though not for a large server base), I'm saying it sounds to me like there's no backup plan if Microsoft makes a mistake that affects your particular systems / software that never gets fixed. This is my fundamental concern about Windows 10 right here.

 

[Jim Kim]

Requirements. CentOS or Ubuntu probably didn't meet their internal requirements.

Do you mean their "crashes and printing errors" requirements? I have it on good authority that CentOS or Ubuntu could have met those requirements, and many more.

 

[dgingeri]

So are you saying you don't have a plan if an update causes a problem that simply doesn't get fixed? On Windows 7, you simply skip that particular update. On Windows 10, you cannot update again if that happens. I'm not saying you've ever had that problem (I have, though not for a large server base), I'm saying it sounds to me like there's no backup plan if Microsoft makes a mistake that affects your particular systems / software that never gets fixed. This is my fundamental concern about Windows 10 right here.

"simply doesn't get fixed"? Never happens. We've never had this issue. We've had things get changed where our web apps don't work and we've had to adjust their programming. That's what our QA environment is for. We simply stalled the updates until the adjustment to the web apps fixes the issue.

The BIG problem is that there are SO MANY developers out there who think that their software is perfect, and they should never have to make adjustments after an update changes a library function. Sometimes, those functions have to get changed in order to close a security hole, and then other software needs to be updated to adjust to that. Our developers don't have that major ego block, so they understand and adjust. No problem. I have worked with many developers who blame Microsoft for their programs not working, even though they were the ones using a non-standard method that wound up with their program being unable to work after an update, and then refuse to update it, or at the very least whine and gripe about needing to update. That's the way the world works. Duh. I despise developers like that.

 

[dgingeri]

Do you mean their "crashes and printing errors" requirements? I have it on good authority that CentOS or Ubuntu could have met those requirements, and many more.

More like a completed and stable OS without needed software developers to finish aspects of the OS. CentOS wouldn't meet that, and Ubuntu would barely meet it. In order to do just about anything with CentOS, you've got to have extensive training and education on how to program in Linux, and not having that can leave CentOS in a very unstable situation. At least most of the OS in Ubuntu is actually finished, most. It can stand on its own, but many apps needs someone with a software development degree in order to configure and install them.

I know this from extensive first hand experience, as a systems admin without a software development degree, over the last 7 years. I've had to work with CentOS, RedHat, SLES, and Ubuntu extensively. Ubuntu server mostly in the last year and a half and the rest over 6 years at my previous job. The more I work with it, the more I realize just how totally half-assed Linux is: half finished and no consistent logic. It takes knowledge of 4-5 different programming languages to finish some of these programs, too, and every single program has their own unique logic on how to use it. Salt, for example, takes knowledge of Yml, Python, it's own command language, and a couple other programming languages in order to make use of it. Graylog, which I'm having to deal with now, has two different languages just to do a freaking search through the logs. Even worse is that most of Graylog is still in development, and the agent software is still in alpha stages. It takes manually configuring a third party program to actually get servers to send the logs to it, requiring knowledge of another programming language to configure either of the agent software packages. It's a horrible OS, and the whole idea behind "open source" is idiocy.

An OS NEEDS to have consistent logic on how to work with things and NEEDS to actually be finished before releasing it.

Average people will never be able to use something without consistent logic, and will never consent to use something that isn't even done yet.

 

[Frobozz]

More like a completed and stable OS without needed software developers to finish aspects of the OS. CentOS wouldn't meet that, and Ubuntu would barely meet it. In order to do just about anything with CentOS, you've got to have extensive training and education on how to program in Linux, and not having that can leave CentOS in a very unstable situation. At least most of the OS in Ubuntu is actually finished, most. It can stand on its own, but many apps needs someone with a software development degree in order to configure and install them.

I know this from extensive first hand experience, as a systems admin without a software development degree, over the last 7 years. I've had to work with CentOS, RedHat, SLES, and Ubuntu extensively. Ubuntu server mostly in the last year and a half and the rest over 6 years at my previous job. The more I work with it, the more I realize just how totally half-assed Linux is: half finished and no consistent logic. It takes knowledge of 4-5 different programming languages to finish some of these programs, too, and every single program has their own unique logic on how to use it. Salt, for example, takes knowledge of Yml, Python, it's own command language, and a couple other programming languages in order to make use of it. Graylog, which I'm having to deal with now, has two different languages just to do a freaking search through the logs. Even worse is that most of Graylog is still in development, and the agent software is still in alpha stages. It takes manually configuring a third party program to actually get servers to send the logs to it, requiring knowledge of another programming language to configure either of the agent software packages. It's a horrible OS, and the whole idea behind "open source" is idiocy.

An OS NEEDS to have consistent logic on how to work with things and NEEDS to actually be finished before releasing it.

Average people will never be able to use something without consistent logic, and will never consent to use something that isn't even done yet.

Were you compiling desktop applications by hand or something?

 

[dgingeri]

Were you compiling desktop applications by hand or something?

You have experience with a desktop Linux app you haven't had to use some exceedingly complex command line thing to install? I haven't yet.

 

[Frobozz]

You have experience with a desktop Linux app you haven't had to use some exceedingly complex command line thing to install? I haven't yet.

Nothing that an average user (since that's where you put the bar) would be doing. Ex. Setting up email or using Office. Now, orchestration and lifecycle management, sure that's a bit more involved, but no more so than the skillset needed to be an effective SCCM admin.

 

[BulletDust]

You have experience with a desktop Linux app you haven't had to use some exceedingly complex command line thing to install? I haven't yet.

As a sysadmin you should know how to do this, it's your job?!

What you're describing is like being a mechanic, that works only on Toyota's....

Software is not that difficult to install under Linux and the OS is by no means half finished - The days of compiling software from scratch under Linux are pretty much in the past now.

 

[dgingeri]

As a sysadmin you should know how to do this, it's your job?!

What you're describing is like being a mechanic, that works only on Toyota's....

Software is not that difficult to install under Linux and the OS is by no means half finished - The days of compiling software from scratch under Linux are pretty much in the past now.

I know how to do a lot, but obviously not everything. I had been a desktop support tech before I got my test lab sysadmin job 7 and a half years ago, and I learned a lot about doing administration for some Linux items while at that job, but I am by no means a programmer, and I should not have to be. Systems administration is just that: administration, meaning caretaking and allocating the resources. It's NOT programming. Linux shit is half finished and requires programming to finish it, which is WRONG for any OS or systems administration. I can keep track of all the systems under my care; create, modify, and delete VMs as needed; create, modify, and delete RAID volumes on a variety of centralized storage; keep track of all the IPs of all the ranges (in the test lab, I had 4 /21 IP ranges to manage) and assign or recycle as needed; keep up on all the services my infrastructure provides, and be on call at any given time to fix any issues that come up on my infrastructure.

This DOES include many services provides by Linux servers, including haproxy, DNS, DHCP, and storage. As much as I hate the massive amounts of extra work Linux demands in these services, I am capable of handling them. That part is massively annoying, demanding ten times as much time to deal with a minority of my systems. They're half finished because the only way to administer them is to manually edit init or config files, which IS only halfway to being done. The vast majority of Linux services are like that: editing a config or init file to actually set it up. Why don't the idiots behind it actually finish it with a decent UI that doesn't allow things like an illegal character to cause the service to fail? (BIND won't take a '_' in a DNS name, and if one is present in one domain file, the whole service crashes and renders it unusable. That's just stupid! Also, in editing a DNS domain file, the admin must increase the version number or it won't replicate the records to other DNS servers properly. Why can't this stupid little thing be automatic?) THAT'S what I mean about half finished.

There are some things that have come up with my recent job, and our CTOs demands of moving away from Microsoft products, that have me dealing with Salt and Graylog, which I just can't get a handle on, all because the idiots behind it couldn't pick just one programming language to use and can't keep to a consistent logic in how their programs work. I don't have the memory capacity to be able to remember 5-6 programming languages in setting this shit up. Both products aren't EVEN half finished, and are buggy, illogical, half-assed shit that deserve to be flushed. Now he's demanding we quit using our current monitoring software and switch to Zabbix for alerts, which is another one of those horrible, half-assed products, no better, and possibly even worse, compared to Graylog. This is NOT what I signed up for in this job, and I'm NOT even CLOSE to getting paid enough to do it. After I leave, he's going to be massively disappointed with having to pay an extra $15,000-20,000 per year to get someone who can deal with the shit he's shoved on us.

 

[BulletDust]

I know how to do a lot, but obviously not everything. I had been a desktop support tech before I got my test lab sysadmin job 7 and a half years ago, and I learned a lot about doing administration for some Linux items while at that job, but I am by no means a programmer, and I should not have to be. Systems administration is just that: administration, meaning caretaking and allocating the resources. It's NOT programming. Linux shit is half finished and requires programming to finish it, which is WRONG for any OS or systems administration. I can keep track of all the systems under my care; create, modify, and delete VMs as needed; create, modify, and delete RAID volumes on a variety of centralized storage; keep track of all the IPs of all the ranges (in the test lab, I had 4 /21 IP ranges to manage) and assign or recycle as needed; keep up on all the services my infrastructure provides, and be on call at any given time to fix any issues that come up on my infrastructure.

This DOES include many services provides by Linux servers, including haproxy, DNS, DHCP, and storage. As much as I hate the massive amounts of extra work Linux demands in these services, I am capable of handling them. That part is massively annoying, demanding ten times as much time to deal with a minority of my systems. They're half finished because the only way to administer them is to manually edit init or config files, which IS only halfway to being done. The vast majority of Linux services are like that: editing a config or init file to actually set it up. Why don't the idiots behind it actually finish it with a decent UI that doesn't allow things like an illegal character to cause the service to fail? (BIND won't take a '_' in a DNS name, and if one is present in one domain file, the whole service crashes and renders it unusable. That's just stupid! Also, in editing a DNS domain file, the admin must increase the version number or it won't replicate the records to other DNS servers properly. Why can't this stupid little thing be automatic?) THAT'S what I mean about half finished.

There are some things that have come up with my recent job, and our CTOs demands of moving away from Microsoft products, that have me dealing with Salt and Graylog, which I just can't get a handle on, all because the idiots behind it couldn't pick just one programming language to use and can't keep to a consistent logic in how their programs work. I don't have the memory capacity to be able to remember 5-6 programming languages in setting this shit up. Both products aren't EVEN half finished, and are buggy, illogical, half-assed shit that deserve to be flushed. Now he's demanding we quit using our current monitoring software and switch to Zabbix for alerts, which is another one of those horrible, half-assed products, no better, and possibly even worse, compared to Graylog. This is NOT what I signed up for in this job, and I'm NOT even CLOSE to getting paid enough to do it. After I leave, he's going to be massively disappointed with having to pay an extra $15,000-20,000 per year to get someone who can deal with the shit he's shoved on us.

You don't need to be a programmer, you need to be sysadmin and be fully proficient in your trade - That means knowing more than just Windows. Shell scripts are about as far as you'll go programming wise and such solutions are used under Windows also.

Of course you're going to defend Windows in corporate environments, it's all you really know. The lack of a GUI by no means indicates the OS is half finished - Your proficiency may be half finished, but not the OS.

Personally, I think the idea of GUI based snap in's to administer a server based OS is laughable.

 

[dgingeri]

You don't need to be a programmer, you need to be sysadmin and be fully proficient in your trade - That means knowing more than just Windows. Shell scripts are about as far as you'll go programming wise and such solutions are used under Windows also.

Of course you're going to defend Windows in corporate environments, it's all you really know. The lack of a GUI by no means indicates the OS is half finished - Your proficiency may be half finished, but not the OS.

Personally, I think the idea of GUI based snap in's to administer a server based OS is laughable.

Then why do I have to worry about Python, YML, and Jinja for Salt? That's FAR more than just scripting languages. It is FAR easier to ANYTHING with SCCM. SCCM is an actual finished product. SCCM also deals with the logs, instead of having to deal with a program where half of it is still in alpha level development, without even half the planned options for the sidecar even in the current version. It can't even be used to encrypt the logs.

You think it is laughable to SAVE TIME and ENSURE your servers aren't going to go down from a single, uncaught typo? You sound like the type that prefers to keep his job using arcane knowledge instead of real skill. I prefer servers that don't crash from a typo.

 

[Frobozz]

I know how to do a lot, but obviously not everything. I had been a desktop support tech before I got my test lab sysadmin job 7 and a half years ago, and I learned a lot about doing administration for some Linux items while at that job, but I am by no means a programmer, and I should not have to be. Systems administration is just that: administration, meaning caretaking and allocating the resources. It's NOT programming. Linux shit is half finished and requires programming to finish it, which is WRONG for any OS or systems administration. I can keep track of all the systems under my care; create, modify, and delete VMs as needed; create, modify, and delete RAID volumes on a variety of centralized storage; keep track of all the IPs of all the ranges (in the test lab, I had 4 /21 IP ranges to manage) and assign or recycle as needed; keep up on all the services my infrastructure provides, and be on call at any given time to fix any issues that come up on my infrastructure.

This DOES include many services provides by Linux servers, including haproxy, DNS, DHCP, and storage. As much as I hate the massive amounts of extra work Linux demands in these services, I am capable of handling them. That part is massively annoying, demanding ten times as much time to deal with a minority of my systems. They're half finished because the only way to administer them is to manually edit init or config files, which IS only halfway to being done. The vast majority of Linux services are like that: editing a config or init file to actually set it up. Why don't the idiots behind it actually finish it with a decent UI that doesn't allow things like an illegal character to cause the service to fail? (BIND won't take a '_' in a DNS name, and if one is present in one domain file, the whole service crashes and renders it unusable. That's just stupid! Also, in editing a DNS domain file, the admin must increase the version number or it won't replicate the records to other DNS servers properly. Why can't this stupid little thing be automatic?) THAT'S what I mean about half finished.

There are some things that have come up with my recent job, and our CTOs demands of moving away from Microsoft products, that have me dealing with Salt and Graylog, which I just can't get a handle on, all because the idiots behind it couldn't pick just one programming language to use and can't keep to a consistent logic in how their programs work. I don't have the memory capacity to be able to remember 5-6 programming languages in setting this shit up. Both products aren't EVEN half finished, and are buggy, illogical, half-assed shit that deserve to be flushed. Now he's demanding we quit using our current monitoring software and switch to Zabbix for alerts, which is another one of those horrible, half-assed products, no better, and possibly even worse, compared to Graylog. This is NOT what I signed up for in this job, and I'm NOT even CLOSE to getting paid enough to do it. After I leave, he's going to be massively disappointed with having to pay an extra $15,000-20,000 per year to get someone who can deal with the shit he's shoved on us.

I think they key is in your last paragraph. You're extraordinarily frustrated with you environment. The work you're doing genuinely sounds interesting, engaging and a great learning opportunity, but I 100% know that deadlines, an ass manager, and general stress can suck those qualities out of a situation in a heartbeat.

Windows has the same situations, just sometimes different. Ex. crappy log files, the 100 ways software is packaged means that you personally have to build wrappers with powershell with all sorts of goofy logic to get SCCM to deploy it correctly. (getting an MSI from someone to put in that thing is a genuine joy)

 

[dgingeri]

I think they key is in your last paragraph. You're extraordinarily frustrated with you environment. The work you're doing genuinely sounds interesting, engaging and a great learning opportunity, but I 100% know that deadlines, an ass manager, and general stress can suck those qualities out of a situation in a heartbeat.

Windows has the same situations, just sometimes different. Ex. crappy log files, the 100 ways software is packaged means that you personally have to build wrappers with powershell with all sorts of goofy logic to get SCCM to deploy it correctly. (getting an MSI from someone to put in that thing is a genuine joy)

Yeah, I will admit that Windows does have crappy logs.

I do, however, refuse to work with programs that have bad installers. Bad installers means likely bad programs. I always evaluate programs before installing them in our systems, and I can, and have, refused programs for being poorly packaged. Every time I've dealt with a poorly packaged program, it also meant dealing with bugs in the program and poor support and/or patching. These usually don't outlast the current OS, and once a new version of Windows comes out, it won't work with it and will never be updated to work with the new Windows. That, to me, is piss-poor programming. (Those usually come form the most egotistical programmers, who think their programs are perfect and they can do no wrong. I hate big egos more than anything in the world, and do my best to deflate them whenever possible.) If people are unwilling to spend the time and money to properly package their program, then they won't spend the time and money to properly develop or support their program, and any business is better off not even trying to use the program.

That's a big difference between me and other admins: I avoid problems rather than try to fix them all the time. I never have to research a crashed server. My servers don't crash.

 

[Frobozz]

Yeah, I will admit that Windows does have crappy logs.

I do, however, refuse to work with programs that have bad installers. Bad installers means likely bad programs. I always evaluate programs before installing them in our systems, and I can, and have, refused programs for being poorly packaged. Every time I've dealt with a poorly packaged program, it also meant dealing with bugs in the program and poor support and/or patching. .

Now THAT must be nice!

I generally get told "let's try our best to make it happen." (edit: and try to share the attitude.)

 

[dgingeri]

Now THAT must be nice!

I generally get told "let's try our best to make it happen." (edit: and try to share the attitude.)

My employers have already dealt with Shoretel's Outlook addon. They know how much bad software can damage a company. Oddly, that was even before I got here. So, they believe me and go along with it. I do my best to follow up with "there is this program that does the same thing and is supported better." Well, that was the way things were before the merger and new CTO.

In my previous job, I told them programs would be bad, simply by looking at how they installed, and they didn't believe me. They tried it, it cost them time and money, and then they started to believe me. By the last 3 years I was there, I could find a program that would do what they wanted, packaged and supported properly, as an alternative to whatever they requested. They followed my lead, and we did not have a single server crash for the last 3 years I was there. Of course, they also only gave me a 1% raise in 4 and a half years, claiming I was at the cap for my position, so I left. They found out pretty quickly why I was worth far more than they were paying me, and are now looking for my third replacement. They're still butt hurt about my leaving, so they refuse to ask me back, but they are having a heck of a time finding anyone worth what I was.

 

[BSmith]

Wow. Just Wow!

Simply put. You are a Windows sys admin, and not a UNIX sys admin. No offense. I have been doing UNIX sys admin work for over 30 years and every rant of yours is just saying, "I am out of my element and have no idea where to turn." I get frustrated when I am out of my element too.

 

[BulletDust]

Wow. Just Wow!

Simply put. You are a Windows sys admin, and not a UNIX sys admin. No offense. I have been doing UNIX sys admin work for over 30 years and every rant of yours is just saying, "I am out of my element and have no idea where to turn." I get frustrated when I am out of my element too.

That's exactly what I'm reading...