BTW, obscure server names aren't merely useless for security, they make your security worse by slowing down incident response times.
"Alert: Unusual SMTP traffic from HQ-MAIL14."
Oh, right, the email guys are rolling out a new Exchange server. No big deal.
"Alert: Unusual SMTP traffic from NYC-FILE03."
*gets on IM* Hey, New York, we're gonna isolate one of your file servers to investigate some weird traffic. You're cool with your people failing back to a different server, right?
"Alert: Unusual SMTP traffic from BOROMIR"
Ahhhhh what the fuck does BOROMIR do!? Where the fuck is it?
every person who doesn't like an acquired taste always seems to think everyone who likes it is faking it. it should be an official fallacy.
BTW, obscure server names aren't merely useless for security, they make your security worse by slowing down incident response times.
"Alert: Unusual SMTP traffic from HQ-MAIL14."
Oh, right, the email guys are rolling out a new Exchange server. No big deal.
"Alert: Unusual SMTP traffic from NYC-FILE03."
*gets on IM* Hey, New York, we're gonna isolate one of your file servers to investigate some weird traffic. You're cool with your people failing back to a different server, right?
"Alert: Unusual SMTP traffic from BOROMIR"
Ahhhhh what the fuck does BOROMIR do!? Where the fuck is it?
Which is my real answer to your previous question.
In an environment with 5-10 servers, absolutely be whimsical and give your servers personality. You can hold all that shit in your head no problem, and as much as some people may argue, whimsy and personality are assets when used appropriately.
In an environment with more than that, don't be an idiot. Use descriptive names.
BTW, obscure server names aren't merely useless for security, they make your security worse by slowing down incident response times.
"Alert: Unusual SMTP traffic from HQ-MAIL14."
Oh, right, the email guys are rolling out a new Exchange server. No big deal.
"Alert: Unusual SMTP traffic from NYC-FILE03."
*gets on IM* Hey, New York, we're gonna isolate one of your file servers to investigate some weird traffic. You're cool with your people failing back to a different server, right?
"Alert: Unusual SMTP traffic from BOROMIR"
Ahhhhh what the fuck does BOROMIR do!? Where the fuck is it?
BTW, obscure server names aren't merely useless for security, they make your security worse by slowing down incident response times.
"Alert: Unusual SMTP traffic from HQ-MAIL14."
Oh, right, the email guys are rolling out a new Exchange server. No big deal.
"Alert: Unusual SMTP traffic from NYC-FILE03."
*gets on IM* Hey, New York, we're gonna isolate one of your file servers to investigate some weird traffic. You're cool with your people failing back to a different server, right?
"Alert: Unusual SMTP traffic from BOROMIR"
Ahhhhh what the fuck does BOROMIR do!? Where the fuck is it?
He always dies.
That's why I only use EDDARD for headless servers.
every person who doesn't like an acquired taste always seems to think everyone who likes it is faking it. it should be an official fallacy.
yeah I think giving cute names to your servers is a holdover from the bad old days
when the server was a thing down in the IT basement, and you only had two or three of them
you loved them like your children, you had to fight for the budget to buy them and have kept them alive far longer than they should have lived
nowadays a server gets spun up in 20 seconds and trashed 5 minutes later
and my first non-desktop job everything was named like [room][row][rack][uposition]
which was excellent for what I did, which was fixing the busted shit cause if it could call home at all at least it told you where it was
probably caused a lot of headaches for anyone without my job tho
oh and for me too if they got moved and the fuckwit who moved 'em didn't rename them
galaxy brain server naming scheme: hostname is asset tag/serial number, if anyone has questions you just enter the serial into the accurate, fast, and detailed asset tracking database you totally have
life's a game that you're bound to lose / like using a hammer to pound in screws
fuck up once and you break your thumb / if you're happy at all then you're god damn dumb
that's right we're on a fucked up cruise / God is dead but at least we have booze
bad things happen, no one knows why / the sun burns out and everyone dies
galaxy brain server naming scheme: hostname is asset tag/serial number, if anyone has questions you just enter the serial into the accurate, fast, and detailed asset tracking database you totally have
That sounds like it'd get annoying quickly.
every person who doesn't like an acquired taste always seems to think everyone who likes it is faking it. it should be an official fallacy.
galaxy brain server naming scheme: hostname is asset tag/serial number, if anyone has questions you just enter the serial into the accurate, fast, and detailed asset tracking database you totally have
That sounds like it'd get annoying quickly.
IF your asset database is accurate it's a dream
if
life's a game that you're bound to lose / like using a hammer to pound in screws
fuck up once and you break your thumb / if you're happy at all then you're god damn dumb
that's right we're on a fucked up cruise / God is dead but at least we have booze
bad things happen, no one knows why / the sun burns out and everyone dies
I had to spend over 20 minutes today explaining how our field engineers cannot do anything about a comcast outage to a CEO who wasn't listening and bitching that the contract says "complete IT care" and his internet didn't work so fix it
galaxy brain server naming scheme: hostname is asset tag/serial number, if anyone has questions you just enter the serial into the accurate, fast, and detailed asset tracking database you totally have
That sounds like it'd get annoying quickly.
Would be great and work perfectly if you had an amazing configuration management system with all data kept super up to date.
Which literally no one has ever or will ever have.
galaxy brain server naming scheme: hostname is asset tag/serial number, if anyone has questions you just enter the serial into the accurate, fast, and detailed asset tracking database you totally have
That sounds like it'd get annoying quickly.
Would be great and work perfectly if you had an amazing configuration management system with all data kept super up to date.
Which literally no one has ever or will ever have.
We use Kaseya to constantly update ConnectWise configurations, and a link to a password vault with a password sync tool. The configs always have accurate info on device name, drive sizes, memory, cpu, IP address, last user to log on, a link you can click to show the current passwords for any local accounts, associated device configs (an esx host has a bundled config associating it with the vcenter server, etc) and all the past service tickets that involved that device.
Sometimes it breaks and doesn't work right, but generally it's pretty up to date and accurate. Assuming whoever set up a given device did all the right configuration for the reporting.
0
Options
Donovan PuppyfuckerA dagger in the dark isworth a thousand swords in the morningRegistered Userregular
galaxy brain server naming scheme: hostname is asset tag/serial number, if anyone has questions you just enter the serial into the accurate, fast, and detailed asset tracking database you totally have
That sounds like it'd get annoying quickly.
Would be great and work perfectly if you had an amazing configuration management system with all data kept super up to date.
Which literally no one has ever or will ever have.
We use Kaseya to constantly update ConnectWise configurations, and a link to a password vault with a password sync tool. The configs always have accurate info on device name, drive sizes, memory, cpu, IP address, last user to log on, a link you can click to show the current passwords for any local accounts, associated device configs (an esx host has a bundled config associating it with the vcenter server, etc) and all the past service tickets that involved that device.
Sometimes it breaks and doesn't work right, but generally it's pretty up to date and accurate. Assuming whoever set up a given device did all the right configuration for the reporting.
I mean, yeah, my environment is automated to shit and back too.
And yet something is always just fucked enough that I can trust nothing.
galaxy brain server naming scheme: hostname is asset tag/serial number, if anyone has questions you just enter the serial into the accurate, fast, and detailed asset tracking database you totally have
That sounds like it'd get annoying quickly.
Would be great and work perfectly if you had an amazing configuration management system with all data kept super up to date.
Which literally no one has ever or will ever have.
Even if you do, I can still see it being annoying in conversations.
"Hey, is one of our VMware hosts down?"
"Yeah, FWH9012W0Q is down. All of its VMs were redistributed onto FWH9012W1L and FWH9013C3R."
If you have a perfect, responsive, high-availability asset database, then it becomes trivial for you to correlate a human memorable name (NYC-VM-04) to a serial number. So there's no advantage I can see to naming the physical host with a serial number.
Also: if you're using Cisco UCS blades, that really doesn't work, because USC configuration profiles can be freely pushed between blades. You upgrade NYC-VM-04 by slotting in a new blade, copying over the configuration profiles, and pulling out the old blade.
Also also: it doesn't work at all for VMs.
every person who doesn't like an acquired taste always seems to think everyone who likes it is faking it. it should be an official fallacy.
galaxy brain server naming scheme: hostname is asset tag/serial number, if anyone has questions you just enter the serial into the accurate, fast, and detailed asset tracking database you totally have
That sounds like it'd get annoying quickly.
Would be great and work perfectly if you had an amazing configuration management system with all data kept super up to date.
Which literally no one has ever or will ever have.
Even if you do, I can still see it being annoying in conversations.
"Hey, is one of our VMware hosts down?"
"Yeah, FWH9012W0Q is down. All of its VMs were redistributed onto FWH9012W1L and FWH9013C3R."
If you have a perfect, responsive, high-availability asset database, then it becomes trivial for you to correlate a human memorable name (NYC-VM-04) to a serial number. So there's no advantage I can see to naming the physical host with a serial number.
Also: if you're using Cisco UCS blades, that really doesn't work, because USC configuration profiles can be freely pushed between blades. You upgrade NYC-VM-04 by slotting in a new blade, copying over the configuration profiles, and pulling out the old blade.
Also also: it doesn't work at all for VMs.
Ooh I have opinions here!
We have a master list of names which contains >1000 words servers can be, which are only used once. So you get "able" and "fitch" and "subsidy" and things like that. It works great - they're instantly memorable and communicable.
Literally every other attempt at this - where people try to get clever and encode information in - has had two things happens - (1) no one can pronounce anything (so bwromswn14 in unusable and immemorable) and (2) someone swears up and down that "blah information never changes" and then proceeds to relocate server racks between floors or something 6 months later, but that doesn't justify changing server names at all,
What I do have is a pretty damn good CLI tool for handling SSH on servers - I can type `tool ssh bwr*` and it'll connect me to any machine which broadly matches that name that we have a record of somewhere. It also works for looking up servers.
So basically: use mnemonic names, and if you're trying to use your hostnames as an asset database then stop doing that and go install this: https://github.com/digitalocean/netbox/ (don't do what we did where I had to prove we should actually use a database because a giant fucking JSON file in Git is unmanageable).
galaxy brain server naming scheme: hostname is asset tag/serial number, if anyone has questions you just enter the serial into the accurate, fast, and detailed asset tracking database you totally have
That sounds like it'd get annoying quickly.
Would be great and work perfectly if you had an amazing configuration management system with all data kept super up to date.
Which literally no one has ever or will ever have.
Even if you do, I can still see it being annoying in conversations.
"Hey, is one of our VMware hosts down?"
"Yeah, FWH9012W0Q is down. All of its VMs were redistributed onto FWH9012W1L and FWH9013C3R."
If you have a perfect, responsive, high-availability asset database, then it becomes trivial for you to correlate a human memorable name (NYC-VM-04) to a serial number. So there's no advantage I can see to naming the physical host with a serial number.
Also: if you're using Cisco UCS blades, that really doesn't work, because USC configuration profiles can be freely pushed between blades. You upgrade NYC-VM-04 by slotting in a new blade, copying over the configuration profiles, and pulling out the old blade.
Also also: it doesn't work at all for VMs.
Ooh I have opinions here!
We have a master list of names which contains >1000 words servers can be, which are only used once. So you get "able" and "fitch" and "subsidy" and things like that. It works great - they're instantly memorable and communicable.
Literally every other attempt at this - where people try to get clever and encode information in - has had two things happens - (1) no one can pronounce anything (so bwromswn14 in unusable and immemorable) and (2) someone swears up and down that "blah information never changes" and then proceeds to relocate server racks between floors or something 6 months later, but that doesn't justify changing server names at all,
What I do have is a pretty damn good CLI tool for handling SSH on servers - I can type `tool ssh bwr*` and it'll connect me to any machine which broadly matches that name that we have a record of somewhere. It also works for looking up servers.
So basically: use mnemonic names, and if you're trying to use your hostnames as an asset database then stop doing that and go install this: https://github.com/digitalocean/netbox/ (don't do what we did where I had to prove we should actually use a database because a giant fucking JSON file in Git is unmanageable).
I'm thinking about getting a UBNT nanoHD to handle the wireless in my home, but my only available mounting point with ethernet is on a wall in the middle of the house. How directional does their stuff tend to be?
I'm thinking about getting a UBNT nanoHD to handle the wireless in my home, but my only available mounting point with ethernet is on a wall in the middle of the house. How directional does their stuff tend to be?
they don't seem to publish radiation patterns for these, but given that they can be wall/ceiling mounted -- I would assume the signal propagation is primarily "down and out." So if you're mounting vertically on a wall, you'll probably get good coverage in the space in front of the AP. If there's a lot of space behind it (i.e. on the other side of the wall), I wouldn't expect to get much in the way of signal there. You're likely to get some, but it isn't primarily radiating that way and it's also going straight into a wall.
There's a definite dead zone directly behind the antenna on all of the other models with a similar shape, so I expect the nanoHD to be like those. My table-mount idea may not be as terrible as I imagined, which is good.
I'm trying to get away from a mesh/extender solution because in my experience the stuff my wife and I own does a terrible job of handing off between APs.
I'd much rather spend $200 on one really good AP (don't need a router) and deal with issues at the fringes than $300+ on a mesh that annoys my wife when she walks upstairs.
I'm trying to get away from a mesh/extender solution because in my experience the stuff my wife and I own does a terrible job of handing off between APs.
I'd much rather spend $200 on one really good AP (don't need a router) and deal with issues at the fringes than $300+ on a mesh that annoys my wife when she walks upstairs.
The old style "wifi extenders" are way different than the modern mesh systems you get from Eero, Google Wifi, etc. The technology is way different/better.
Posts
the "no true scotch man" fallacy.
the "no true scotch man" fallacy.
[edit]
Spelling is hard.
"Alert: Unusual SMTP traffic from HQ-MAIL14."
Oh, right, the email guys are rolling out a new Exchange server. No big deal.
"Alert: Unusual SMTP traffic from NYC-FILE03."
*gets on IM* Hey, New York, we're gonna isolate one of your file servers to investigate some weird traffic. You're cool with your people failing back to a different server, right?
"Alert: Unusual SMTP traffic from BOROMIR"
Ahhhhh what the fuck does BOROMIR do!? Where the fuck is it?
the "no true scotch man" fallacy.
Which is my real answer to your previous question.
In an environment with 5-10 servers, absolutely be whimsical and give your servers personality. You can hold all that shit in your head no problem, and as much as some people may argue, whimsy and personality are assets when used appropriately.
In an environment with more than that, don't be an idiot. Use descriptive names.
He always dies.
That's why I only use EDDARD for headless servers.
the "no true scotch man" fallacy.
Maybe. . .
XBL:Phenyhelm - 3DS:Phenyhelm
I used to do Mass Effect names.
These days it's mostly Prey.
the "no true scotch man" fallacy.
when the server was a thing down in the IT basement, and you only had two or three of them
you loved them like your children, you had to fight for the budget to buy them and have kept them alive far longer than they should have lived
nowadays a server gets spun up in 20 seconds and trashed 5 minutes later
and my first non-desktop job everything was named like [room][row][rack][uposition]
which was excellent for what I did, which was fixing the busted shit cause if it could call home at all at least it told you where it was
probably caused a lot of headaches for anyone without my job tho
oh and for me too if they got moved and the fuckwit who moved 'em didn't rename them
galaxy brain server naming scheme: hostname is asset tag/serial number, if anyone has questions you just enter the serial into the accurate, fast, and detailed asset tracking database you totally have
fuck up once and you break your thumb / if you're happy at all then you're god damn dumb
that's right we're on a fucked up cruise / God is dead but at least we have booze
bad things happen, no one knows why / the sun burns out and everyone dies
My home computers are all named for what sort of detriment they ultimately render upon my life.
"TimeRuiner" - My desktop gaming PC
"StudyRuiner" - My school laptop
"SanityRuiner" - My wife's PC
"SleepRuiner" - A little laptop on my nightstand
That sounds like it'd get annoying quickly.
the "no true scotch man" fallacy.
the "no true scotch man" fallacy.
IF your asset database is accurate it's a dream
if
fuck up once and you break your thumb / if you're happy at all then you're god damn dumb
that's right we're on a fucked up cruise / God is dead but at least we have booze
bad things happen, no one knows why / the sun burns out and everyone dies
One of my very favorite multi-use phrases is "this is only temporary unless it works".
I need a raise
It runs Prometheus.
Would be great and work perfectly if you had an amazing configuration management system with all data kept super up to date.
Which literally no one has ever or will ever have.
Working as intended.
We use Kaseya to constantly update ConnectWise configurations, and a link to a password vault with a password sync tool. The configs always have accurate info on device name, drive sizes, memory, cpu, IP address, last user to log on, a link you can click to show the current passwords for any local accounts, associated device configs (an esx host has a bundled config associating it with the vcenter server, etc) and all the past service tickets that involved that device.
Sometimes it breaks and doesn't work right, but generally it's pretty up to date and accurate. Assuming whoever set up a given device did all the right configuration for the reporting.
My PC is named Nikolai because I felt like it. Also Tesla was a BAWSS.
I'd have named it Laika but that's what I'm calling my dog when I get it.
I mean, yeah, my environment is automated to shit and back too.
And yet something is always just fucked enough that I can trust nothing.
Even if you do, I can still see it being annoying in conversations.
"Hey, is one of our VMware hosts down?"
"Yeah, FWH9012W0Q is down. All of its VMs were redistributed onto FWH9012W1L and FWH9013C3R."
If you have a perfect, responsive, high-availability asset database, then it becomes trivial for you to correlate a human memorable name (NYC-VM-04) to a serial number. So there's no advantage I can see to naming the physical host with a serial number.
Also: if you're using Cisco UCS blades, that really doesn't work, because USC configuration profiles can be freely pushed between blades. You upgrade NYC-VM-04 by slotting in a new blade, copying over the configuration profiles, and pulling out the old blade.
Also also: it doesn't work at all for VMs.
the "no true scotch man" fallacy.
Ooh I have opinions here!
We have a master list of names which contains >1000 words servers can be, which are only used once. So you get "able" and "fitch" and "subsidy" and things like that. It works great - they're instantly memorable and communicable.
Literally every other attempt at this - where people try to get clever and encode information in - has had two things happens - (1) no one can pronounce anything (so bwromswn14 in unusable and immemorable) and (2) someone swears up and down that "blah information never changes" and then proceeds to relocate server racks between floors or something 6 months later, but that doesn't justify changing server names at all,
What I do have is a pretty damn good CLI tool for handling SSH on servers - I can type `tool ssh bwr*` and it'll connect me to any machine which broadly matches that name that we have a record of somewhere. It also works for looking up servers.
So basically: use mnemonic names, and if you're trying to use your hostnames as an asset database then stop doing that and go install this: https://github.com/digitalocean/netbox/ (don't do what we did where I had to prove we should actually use a database because a giant fucking JSON file in Git is unmanageable).
I can get behind all of this.
the "no true scotch man" fallacy.
they don't seem to publish radiation patterns for these, but given that they can be wall/ceiling mounted -- I would assume the signal propagation is primarily "down and out." So if you're mounting vertically on a wall, you'll probably get good coverage in the space in front of the AP. If there's a lot of space behind it (i.e. on the other side of the wall), I wouldn't expect to get much in the way of signal there. You're likely to get some, but it isn't primarily radiating that way and it's also going straight into a wall.
With the way my house is laid out I might actually be better off laying it on its "back" on the table where my fiber drop comes in...
There's a definite dead zone directly behind the antenna on all of the other models with a similar shape, so I expect the nanoHD to be like those. My table-mount idea may not be as terrible as I imagined, which is good.
I'd much rather spend $200 on one really good AP (don't need a router) and deal with issues at the fringes than $300+ on a mesh that annoys my wife when she walks upstairs.
You can also just crank down the power of the stuff at the fringes so that it doesn't reach much past the area it's in.
The old style "wifi extenders" are way different than the modern mesh systems you get from Eero, Google Wifi, etc. The technology is way different/better.