"safe removal" for usb in fedora 17 removed?

[robertdaleweir]

Originally Posted by wrong

I was toying around and found this :

If I write

or


both replies with




So, I cd-ed to /dev/ and write ls sd* and found few outputs, and two importants are

sdb sdb1

so, I tried with this command


everything went fine, and next thing that I done was


and my disk stoped

So, if some of you don't like to eject disk from "Files" and then go to terminal and detach, you can do what I described.

I suppose that if you have more partitions on disk, you shuld unmount all of them and then try to detach.

Hi wrong
As long as the partition is unmounted there should be no data loss, ie it would be flushed to disk or whatever. I had thought that USB was Hot Swappable because of the fact that the Power Pins engage 'first' on insert and 'last' on withdrawal, ensuring power before data and vice versa.
Hence the power off aspect is purely one of Energy Usage (possibly drive wear and tear) and not one of Data Loss. Hence from a data integrity point of view Unmounting should be the only worry when you disconnnect. The Eject should handle that no problem. The power is not important unless something automatically keeps remounting it when the power is left on , then you have a possible issue.
Cheers...

[wrong]

Originally Posted by robertdaleweir

Hi wrong
As long as the partition is unmounted there should be no data loss, ie it would be flushed to disk or whatever. I had thought that USB was Hot Swappable because of the fact that the Power Pins engage 'first' on insert and 'last' on withdrawal, ensuring power before data and vice versa.
Hence the power off aspect is purely one of Energy Usage (possibly drive wear and tear) and not one of Data Loss. Hence from a data integrity point of view Unmounting should be the only worry when you disconnnect. The Eject should handle that no problem. The power is not important unless something automatically keeps remounting it when the power is left on , then you have a possible issue.
Cheers...

Hi,

the only reason why I want to cut the power off is that there is no reason to keep my hdd spinning while I have nothing to do. So, unmounting/ejecting doesn't do the trick, i detach him so it keeps calm. And I found this thread and thought it would be interesting to share those info with you.

[DBelton]

I was thinking that I saw a bug someplace where the cache wasn't flushed properly and data corruption occurred. That possibly could be the reason they removed the option (hopefully temporarily until the bug is fixed)

But, most external USB drives will spin the drive down after a certain period of inactivity, so your drive isn't constantly spinning anyway. (That caused issues back around F10-11 where the drive would spin down, but linux thought it was still ready and would start sending data instead of waiting for it to spin back up)

[secipolla]

Originally Posted by DBelton

I was thinking that I saw a bug someplace where the cache wasn't flushed properly and data corruption occurred. That possibly could be the reason they removed the option (hopefully temporarily until the bug is fixed)

You mean on unmount?
Then it wouldn't make any difference if it's detached or not.

[robertdaleweir]

Originally Posted by wrong

Hi,

the only reason why I want to cut the power off is that there is no reason to keep my hdd spinning while I have nothing to do. So, unmounting/ejecting doesn't do the trick, i detach him so it keeps calm. And I found this thread and thought it would be interesting to share those info with you.

Hi wrong (love your user name)
No problem. Its all good. The more information out there the better. Have a great day...

[BillBauman]

Hey, guys, I've been experiencing this on F16 and F17, both x64. I can confirm this is a USB 3.0 issue. I can safely detach (power down) a physical device plugged into a USB 2.0 port, but not a device on a USB 3.0 port. This has been a problem for me with every USB 3.0 device I've tried since the inception of its support in Linux. I consider it a relatively significant bug.

I'm also extremely annoyed that the short-sighted Gnome community made another stupid decision in removing the "Safely Remove" / detach / power down feature from the GUI. I understand that data is quiesced by unmounting, but unplugging a device that is actively powered is not recommended in any practice of electrical engineering.

Hot-swappable/pluggable drives 101: Always power down before removal. Always. There seems to be a significant amount of ignorance in the software development community with regard to hardware management.

[MauserM98]

In KDE safe removal is present and working. I have no USB 3 devices so I have not tested that yet.
Why are you complaining about Gnome by the way? Just leave.. I did.

[BillBauman]

Originally Posted by MauserM98

In KDE safe removal is present and working. I have no USB 3 devices so I have not tested that yet.
Why are you complaining about Gnome by the way? Just leave.. I did.

I can "safe remove" with udisks --detach at the CLI, so I'm okay with that. The bug with USB 3.0 will probably become more apparent as the ports/devices become more prevalent. It might be limited to particular chipsets for all I know, I've only tested it on Thinkpad T520 and T420S, both based on the same USB 3.0 platform.

As far as Gnome complaining, I hear you, and I tried, but there are just a few deal breakers in KDE for me. I also tried LXDE and XFCE. When it came down to it, Gnome 3 + Cinnamon is as close to the user experience I prefer as I'm able to achieve. Honestly Gnome 3 truly is the "unholy hell" that Linus called it. Cinnamon is the only thing that made it viable for me to upgrade past F14. Were it not for Cinnamon, I'd be back on F14, but newer hardware really benefits from the newer kernel.

[deanej]

Isn't the point of hot swappable to be that you don't have to power anything down? In Windows XP I would just yank jump drives out without issue (though Windows 7 will complain so I stopped). Safe removal is all about stopping/completing I/O so you don't have data loss.

[BillBauman]

Originally Posted by deanej

Isn't the point of hot swappable to be that you don't have to power anything down? In Windows XP I would just yank jump drives out without issue (though Windows 7 will complain so I stopped). Safe removal is all about stopping/completing I/O so you don't have data loss.

The answer is "sort of". The reality is there's no such thing as "hot swappable", there is, and there isn't. Any time you have an electrical circuit connected, if you simple pull it apart, you run the risk of arcing the circuit and possible damaging one or both of the components previously attached to it. That chance is very slim, and goes down with the amount of amperage in the circuit, in general, as well as goes down depending on the sensitivity of the devices.

So, most of the time, you'll have no problems. Sometimes, you might. Some USB drives just "go dead" after getting pulled, it could just be the cells died, but it's also very likely that it got jolted out of existence when it was yanked, as you say.

Unmounting / ejecting simply quiesces data, meaning, the data flow is stopped. It also flushes the cache prior to that, obviously. Safely removing / detaching / powering down, should disconnect the electrical circuit such that it is now safe to remove without the risk (obviously nothing in life is 100% risk free) of damaging either device.

[robertdaleweir]

Originally Posted by BillBauman

The answer is "sort of". The reality is there's no such thing as "hot swappable", there is, and there isn't. Any time you have an electrical circuit connected, if you simple pull it apart, you run the risk of arcing the circuit and possible damaging one or both of the components previously attached to it. That chance is very slim, and goes down with the amount of amperage in the circuit, in general, as well as goes down depending on the sensitivity of the devices.

So, most of the time, you'll have no problems. Sometimes, you might. Some USB drives just "go dead" after getting pulled, it could just be the cells died, but it's also very likely that it got jolted out of existence when it was yanked, as you say.

Unmounting / ejecting simply quiesces data, meaning, the data flow is stopped. It also flushes the cache prior to that, obviously. Safely removing / detaching / powering down, should disconnect the electrical circuit such that it is now safe to remove without the risk (obviously nothing in life is 100% risk free) of damaging either device.

Hi BillBauman
The way the USB connector is designed, as far as I can tell, the pins for power and data are different lengths and as such the Power Connects first, then the data. Likewise it disconnects last upon removal. We are talking about 5V DC, after all. Upon examining a USB A type connector, you will see that the Pins on the out side of the unit are longer. As such they connect electrically long before the two middle Pins, which are a differential data signal pair. I have never measured the pins, but on quick observation it appears that the Power Pin 1 and the GRD Pin 4 are at least 1/8 inch longer than the inner pins. This distance is large in electrical terms (at 5volts DC and Milliamps of draw), given that the insertion is done relatively slowly.
I believe that USB connections are Hot Swappable and the Data Loss is the only real issue. As long as the data is flushed, anything else that happens to a drive that may have been powered down by removing a USB connector is purely coincidence, in my opinion.
The pinout of a USB connector is attached as a pdf, for reference.

[BillBauman]

Originally Posted by robertdaleweir

Hi BillBauman
The way the USB connector is designed, as far as I can tell, the pins for power and data are different lengths and as such the Power Connects first, then the data. Likewise it disconnects last upon removal. We are talking about 5V DC, after all. Upon examining a USB A type connector, you will see that the Pins on the out side of the unit are longer. As such they connect electrically long before the two middle Pins, which are a differential data signal pair. I have never measured the pins, but on quick observation it appears that the Power Pin 1 and the GRD Pin 4 are at least 1/8 inch longer than the inner pins. This distance is large in electrical terms (at 5volts DC and Milliamps of draw), given that the insertion is done relatively slowly.
I believe that USB connections are Hot Swappable and the Data Loss is the only real issue. As long as the data is flushed, anything else that happens to a drive that may have been powered down by removing a USB connector is purely coincidence, in my opinion.
The pinout of a USB connector is attached as a pdf, for reference.

I don't disagree with your assessment of the USB connection design at all. But simply because the data pins aren't connected doesn't negate the possibilities of what can happen when you yank an active electrical circuit apart. In an excessively unscientific, anecdotal, personal analysis, I can say that I have never had a USB device fail when I have consistently powered it down before detachment. I have had, to my memory, devices fail prior to my paying attention to such things. Clearly, that does not universally support what I'm saying, however, to simply remove or not have the capability is, to me, assinine. Isn't Linux supposed to be more configurable, user-selectable, greater optionable than Windows; not less?

Anyway, I get your point. For the most part, flushing and quiescing the data should prevent most potential loss of anything, but I would still far, far prefer to be able to power off the device prior to disconnected the circuit. Call me old school.

[ddan]

Greetings,

I have been running Fedora 17 for about 3 weeks. I have had 2 corrupted backups. I have been with Fedora since about Fedora 6 and I have never had a corrupted backup until Fedora 17. I have several backup discs, some are USB 3.0, USB 2.0, some have an activity light and some don't. Often when I unmount it says to the drive is Busy. There is nothing using the drive. I have to say "unmount anyway". The drives that have an activity light when an unmount is done the activity continues to blink for up to 30 seconds. But there are drives with no activity light, so it hard to know when to disconnect the drive.

With the "safe removal" all data was written to the drive and when it said it was safe to remove it was safe to remove.

I did find a post about #eject but finding the /dev/?? seems difficult. The only program that I found that converted the label name to /dev/?? is gparted. When I enter eject /dev/sdg1 it says "Can't open device".

I'm not sure what to do about Fedora 17. I have to have reliable backups.

Has anyone found a solution?

ddan

[ddan]

Greetings,

I found that I had to be su for the eject command to work. However the drive doesn't power down.

My biggest problem is the unmount fails. It says it is busy. I only use the external drives for backup and there is not reason for the drive to be busy. They were never busy in F16. This seems to happen mostly with drives that have two partitions. I have to say "Unmount Anyway".

Even on drives that do unmount, there is activity on the drive for maybe 30 seconds after the unmont. If the drive doesn't have an activity light, it is hard to tell when the data is finished being written.

This change doesn't seem to have been thought out well!

Does anyone have a solution? F16 is about to enter EOL.

ddan