The Software RAID Inflection Point
-
@travisdh1 said in The Software RAID Inflection Point:
@momurda said in The Software RAID Inflection Point:
If the md authors would make some sort of blinky light function that would be sweet and solve this issue. mdadm -blink /dev/sde
How do you do that for cases that only have a single HDD light?
Or none.
-
@dafyre said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@dafyre said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@Dashrender said in The Software RAID Inflection Point:
@travisdh1 said in The Software RAID Inflection Point:
@FATeknollogee said in The Software RAID Inflection Point:
@Dashrender said in The Software RAID Inflection Point:
Because the software RAID would rarely, if ever, be aware of the type of hardware it's it, i.e. a HP server, a Dell server, etc - it probably has no way of mapping a drive to a specific port on the chassis and lighting the lights, etc - the things needed to make Blind Swap work.
In today's world with all the smart people we have, no one can write software that can map drives to a chassis?
If a hardware vendor wanted to, they easily could, but it would mean seriously hampering the number of options available on a general purpose server platform. NAS vendors do this all the time.
NAS vendors do it because they control all the hardware. Sure, Dell could do it for their system, and HP for theirs, etc - but why? They'd rather sell you a RAID card.
Actually they can't, because they don't know the use case. It's not about controlling all the hardware, it's about controlling everything from the disk to the application.
What do you mean that they don't know the use case...?
Does it matter? We just want a way to flash the "help me" light on a dying/dead disk, lol.
Would that not be something controllable by the firmware of the HD?
That's not what blind swap does.
No, it doesn't. But if you know which drive is failing, that makes it easier to swap the correct drive.
True, but we already have that feature.
-
Here is the software RAID blink command for Solaris:
http://prefetch.net/blog/index.php/2007/03/13/locating-disk-drives-in-a-sea-of-a5200s/
-
And the third entry here shows how to do this with Linux, with or without MD RAID.
http://serverfault.com/questions/64239/physically-identify-the-failed-hard-drive
-
@scottalanmiller said in The Software RAID Inflection Point:
And the third entry here shows how to do this with Linux, with or without MD RAID.
http://serverfault.com/questions/64239/physically-identify-the-failed-hard-drive
That's what we were looking for. Why couldn't something like that be integrated with mdraid, much as momurda suggested?
When mdraid detects a failed drive, kick off a script that fires off smartctl or ledctl or something?
-
@dafyre said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
And the third entry here shows how to do this with Linux, with or without MD RAID.
http://serverfault.com/questions/64239/physically-identify-the-failed-hard-drive
That's what we were looking for. Why couldn't something like that be integrated with mdraid, much as momurda suggested?
But it is already. It's all there.
-
@dafyre said in The Software RAID Inflection Point:
When mdraid detects a failed drive, kick off a script that fires off smartctl or ledctl or something?
To just signal the light?
-
@scottalanmiller said in The Software RAID Inflection Point:
@dafyre said in The Software RAID Inflection Point:
When mdraid detects a failed drive, kick off a script that fires off smartctl or ledctl or something?
To just signal the light?
Boy I would hope not - I think @dafyre is asking - why doesn't mdRAID have blind swap? What prevents it from kicking scripts off that light the light and demount the drive in preparation of being replaced?
-
@Dashrender said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@dafyre said in The Software RAID Inflection Point:
When mdraid detects a failed drive, kick off a script that fires off smartctl or ledctl or something?
To just signal the light?
Boy I would hope not - I think @dafyre is asking - why doesn't mdRAID have blind swap? What prevents it from kicking scripts off that light the light and demount the drive in preparation of being replaced?
This is exactly what I am asking.
-
@Dashrender said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@dafyre said in The Software RAID Inflection Point:
When mdraid detects a failed drive, kick off a script that fires off smartctl or ledctl or something?
To just signal the light?
Boy I would hope not - I think @dafyre is asking - why doesn't mdRAID have blind swap? What prevents it from kicking scripts off that light the light and demount the drive in preparation of being replaced?
Well, showing some lights is "easy", you just use the tools that are already there. What stops it from replacing the drive is what we keep saying... it doesn't have enough info to know what to do when a disk is replaced. It's a general purpose system, not a pre-defined NAS. Scripted MD is what almost all major NAS use, so obviously it works. But you can't just do that in the OS.
-
@scottalanmiller said in The Software RAID Inflection Point:
@Dashrender said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@dafyre said in The Software RAID Inflection Point:
When mdraid detects a failed drive, kick off a script that fires off smartctl or ledctl or something?
To just signal the light?
Boy I would hope not - I think @dafyre is asking - why doesn't mdRAID have blind swap? What prevents it from kicking scripts off that light the light and demount the drive in preparation of being replaced?
Well, showing some lights is "easy", you just use the tools that are already there. What stops it from replacing the drive is what we keep saying... it doesn't have enough info to know what to do when a disk is replaced. It's a general purpose system, not a pre-defined NAS. Scripted MD is what almost all major NAS use, so obviously it works. But you can't just do that in the OS.
Now I get what Scott is saying... The NAS systems all have scripted checks for when you replace a drive in the system and such... Basically, the same commands that we would have to run by hand on a stock MD system... We should figure out how to do all this for ourselves in mdraid, lol.
-
@dafyre said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@Dashrender said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@dafyre said in The Software RAID Inflection Point:
When mdraid detects a failed drive, kick off a script that fires off smartctl or ledctl or something?
To just signal the light?
Boy I would hope not - I think @dafyre is asking - why doesn't mdRAID have blind swap? What prevents it from kicking scripts off that light the light and demount the drive in preparation of being replaced?
Well, showing some lights is "easy", you just use the tools that are already there. What stops it from replacing the drive is what we keep saying... it doesn't have enough info to know what to do when a disk is replaced. It's a general purpose system, not a pre-defined NAS. Scripted MD is what almost all major NAS use, so obviously it works. But you can't just do that in the OS.
Now I get what Scott is saying... The NAS systems all have scripted checks for when you replace a drive in the system and such... Basically, the same commands that we would have to run by hand on a stock MD system... We should figure out how to do all this for ourselves in mdraid, lol.
Right, because you can choose your use case, but MD cannot do that on its own without significantly curtailing its functionality. MD offers way more options than a NAS device does, that's part of its power. If it did everything automatically, you'd give a lot of that up.
You could make an add on application that does lights and replacements automatically, of course, and I think in defining all of the things to detect and how to react you'd notice tonnes of "assumptions" that you have to make to make it all work.
-
@scottalanmiller said in The Software RAID Inflection Point:
But you can't just do that in the OS.
why not?
-
@scottalanmiller said in The Software RAID Inflection Point:
@dafyre said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@Dashrender said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@dafyre said in The Software RAID Inflection Point:
When mdraid detects a failed drive, kick off a script that fires off smartctl or ledctl or something?
To just signal the light?
Boy I would hope not - I think @dafyre is asking - why doesn't mdRAID have blind swap? What prevents it from kicking scripts off that light the light and demount the drive in preparation of being replaced?
Well, showing some lights is "easy", you just use the tools that are already there. What stops it from replacing the drive is what we keep saying... it doesn't have enough info to know what to do when a disk is replaced. It's a general purpose system, not a pre-defined NAS. Scripted MD is what almost all major NAS use, so obviously it works. But you can't just do that in the OS.
Now I get what Scott is saying... The NAS systems all have scripted checks for when you replace a drive in the system and such... Basically, the same commands that we would have to run by hand on a stock MD system... We should figure out how to do all this for ourselves in mdraid, lol.
Right, because you can choose your use case, but MD cannot do that on its own without significantly curtailing its functionality. MD offers way more options than a NAS device does, that's part of its power. If it did everything automatically, you'd give a lot of that up.
You could make an add on application that does lights and replacements automatically, of course, and I think in defining all of the things to detect and how to react you'd notice tonnes of "assumptions" that you have to make to make it all work.
But the MD guys would make setup scripts that ask the builder all those questions so that MD is setup as desired. Not unlike using a RAID card.
-
Or does a RAID card something else that I'm missing.
-
@Dashrender said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
But you can't just do that in the OS.
why not?
Because you don't know how it is being used. General purpose, remember?
-
@Dashrender said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@dafyre said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@Dashrender said in The Software RAID Inflection Point:
@scottalanmiller said in The Software RAID Inflection Point:
@dafyre said in The Software RAID Inflection Point:
When mdraid detects a failed drive, kick off a script that fires off smartctl or ledctl or something?
To just signal the light?
Boy I would hope not - I think @dafyre is asking - why doesn't mdRAID have blind swap? What prevents it from kicking scripts off that light the light and demount the drive in preparation of being replaced?
Well, showing some lights is "easy", you just use the tools that are already there. What stops it from replacing the drive is what we keep saying... it doesn't have enough info to know what to do when a disk is replaced. It's a general purpose system, not a pre-defined NAS. Scripted MD is what almost all major NAS use, so obviously it works. But you can't just do that in the OS.
Now I get what Scott is saying... The NAS systems all have scripted checks for when you replace a drive in the system and such... Basically, the same commands that we would have to run by hand on a stock MD system... We should figure out how to do all this for ourselves in mdraid, lol.
Right, because you can choose your use case, but MD cannot do that on its own without significantly curtailing its functionality. MD offers way more options than a NAS device does, that's part of its power. If it did everything automatically, you'd give a lot of that up.
You could make an add on application that does lights and replacements automatically, of course, and I think in defining all of the things to detect and how to react you'd notice tonnes of "assumptions" that you have to make to make it all work.
But the MD guys would make setup scripts that ask the builder all those questions so that MD is setup as desired. Not unlike using a RAID card.
And then it would be limited in the same way, but even moreso, because it would have access to far more things.
-
@Dashrender said in The Software RAID Inflection Point:
Or does a RAID card something else that I'm missing.
RAID card assumes you'll only attach things to it that you want to work in a specific way. MD can't do that.
-
@scottalanmiller said in The Software RAID Inflection Point:
@Dashrender said in The Software RAID Inflection Point:
Or does a RAID card something else that I'm missing.
RAID card assumes you'll only attach things to it that you want to work in a specific way. MD can't do that.
Please help the feable with examples of things that would work in a different way, and why you would expect that to happen in a MD setup, but not in a RAID controller setup.
-
@scottalanmiller said in The Software RAID Inflection Point:
@Dashrender said in The Software RAID Inflection Point:
Or does a RAID card something else that I'm missing.
RAID card assumes you'll only attach things to it that you want to work in a specific way. MD can't do that.
MD Relies on us to tell it how to operate. Part of that would be setting up the blinking lights and all of that too.