Depends on how you mount the device. You can mount a device so that it does not buffer like that. Such as at the end of a USB stick write. Whether that is a good thing or not, I'm still deciding. It has not given me an problems yet.

1 members found this post helpful.

Hard drives are full embedded computers these days with all the disadvantages that entails. I notice that many of the newer devices keep running for at least a few seconds even after being unmounted and ejected from the system even when sync has been used. The system shows the device has detached but the LED on the device blinks for a few more seconds and that data loss entails if it is powered off by force first.

Many disk utilities now mix and match sectors, GB, and GiB in a most confusing manner making it very hard to estimate size. It does not matter much except in some edge cases, such as swap or mirroring, etc. These random changes in units bit me recently on a fresh installation where I had misallocated a few kB too little space on a swap partition such that the "suspend to RAM" option would lock the system fatally it time it was invoked. So what I learned is that the graphical disk utilities, such as in the installer or on the live images, don't give enough information (or use consistent units) to allow fine tuning of the partition sizes.

Care to share how you do that?
I suppose this is a specific option in the fstab or mount command.

That the IPv6 address of localhost is ::1.
And that anything listening on “l o c a l h o s t” had better be equipped for IPv6, these days.

Not true I don't use ipv6 and apache works fine, as does calibre server minidlna etc etc, try not to make sweeping statments like this as it can be off putting for noobs

2 members found this post helpful.

Just as a little FYI, if I enable ipv6 on my TV ( a low end LG smart TV ), the inbuilt netflix app wont connect so not even the big company's are fully using v6 yet.

@Keith,
This is not necessarily true.
The problem could also come from your router or your ISP's infrastructure,
either at the IP level or the DNS level.

probably not though as some of the on demand apps work fine with opv6 selected, just a a netflix oddity probably

http://cht.sh is cool! https://www.linuxquestions.org/quest...3/#post6089690



Sadly, LinuxJourney.com not working since Dec: https://github.com/cindyq/linuxjourn...ent-1370376253

Google returns this first result: notes.siira.io but it seems to be a non-working 'scrape' of just the cover page.

::1 and 127.0.0.1 are most certainly home for system DNS. i learned too use systemd as a DNS caching server it works very well on Xubuntu ...

I have been mounting my usb "stuff" for a while now with
And that is for NTFS and FAT devices. Hard drives and usb sticks. You don't have to coddle ext4 devices.

They write all at once when mounted like that, when the copy process is done, it's done. No waiting for 10-15 seconds for the buffer to write to stick. Now, whether that is a good thing or not, I'm still watching it. I haven't had any corrupt data so far. There are also other mount flags that will affect how a device is written to.

Yup, a NTFS volumne will certainly do that.

I use the NTFS3 that is in the kernel, dropped ntfs-3g a good while back. But, I am not sure that NTFS3 may still have a bug or two. Knowing how picky a NTFS is, I always treat them with special care. A couple of weeks ago, I had a NTFS external HD, that I didn't do anything wrong to. I'm careful with them. It started acting weird out of the blue. So, I stuck it into a win10 machine, run its disk checker on it, something wrong, let windows fix it, problem gone. If NTFS weren't so handy for everything to access, windows, linux, BSD, smart bluray players etc, I would stop using it.

I'm using arch, and for the last year or so, every new kernel brings a different behavior writing to usb devices. One kernel writes smooth and steady, the next kernel writes like a roller coaster. Or... I don't know what else to blame for that.

They are reading, caching, writing differently.

You can even get a different result by making a RAM drive, copying a big file up into it, then writing from it to an external usb device. And I don't mean faster speed. I would expect reading from RAM to be faster that reading from a hard drive. I'm referring to a steady write process. Such as:
That will write at a steady speed, limiting factor is USB bus and USB device speed.

Whereas this is all over the place.(Depends on Kernel)
This month that is, next months update, which brings a new kernel, may by altogether different. And NTFS volumes are the worst. EXT4 doesn't have any problems.

Anyway, that's my 2 cents of what I've seen for the last year or so, with the different kernels. No data loss, just weird usb writing.

If someone has any better insight on this, go ahead and post.

On thursday the 19th, I learned to modify GRUB "gfxterm" theme to display #000 text on #000 background.
Meaning; having multiple menu entries and infinite "wait", it now expects multiple "arrow down" and "enter" to boot anything.
So if I put entry 7 as default, booting the 1st entry requires 6 "arrow up", and since it does not display a thing; it works kinda like a sanity check.

Do the numbers have a special meaning as regards sanity?

I know that it is true that one service stopped working because someone had chosen to modify something in the way I described. This very developer may be especially mad, but if the service is important to me, I cannot care about that.

It could be interesting to have a statistic of some kind, showing how computer users divide into those who take decisions empirically, those who would not move a finger before they haven't seen the code and ... whatever remains. I am quite obsessed with not caring anyway.

Sure they do, booting the thing and not knowing it requires 6 arrows down is futile. And GRUB could have a LOT of entries.
It's basically a sanity check for the one booting the machine, not the kind of software sanity check you would expect by its definition.