Planned downtime: Account format migration

[Gauteamus]

"Platitudibus in lingua altera optime audient." -Ovid

[martenx]

Bottleneck will rise again...

[krikke93]

Bottleneck will rise again...

[goldfighter]

Bottleneck will rise again...

[loftar]

Did you measure how much head repostioning they have to do ? This is more strain on HHD than data transfer alone.

The "70% idle" measure was supposed to be an indication that they are seeking far from constantly. Other than that, I can't say I know how to measure cumulative head travel distance or any other relevant measure.

However, it should be said that if the disks were strained by seeking, I'd expect the SMART "Raw Seek Error Rate" attribute to show that by detereorating over time, but the only thing I have observed with the disks over time is merely that they start getting reallocated sectors, which I, for one, interpret rather as a sign of media failure. I've no clue what would cause it, however.

[g1real]

However, it should be said that if the disks were strained by seeking, I'd expect the SMART "Raw Seek Error Rate" attribute to show that by detereorating over time, but the only thing I have observed with the disks over time is merely that they start getting reallocated sectors, which I, for one, interpret rather as a sign of media failure. I've no clue what would cause it, however.

I would like to thank you mentioning reallocated sectors, I found out one of my HDDs is starting to fail too.

Although I think it's a natural thing. I've run quite some stuff on mine, including many voxel-based game servers, over the 3,5 years I had it. And as you said, you had HDDs replaced that were already used, so it's no surprise. In the end that SDD option might be interesting, as they would show no mechanical failure in the least, and getting a decent sized one can potentially last much much longer - I don't know how much data you write daily, but I believe you said somewhere to 1-2 mb/s, wether that is bits or bytes you didn't really make clear, but I'm assuming bytes. You would end up with 168 GB per day written, with the current limitations of roughly 2500 writes before an SSD is starting to shit itself and a 256 GB drive, you would last more than 8 years.

Anyway, maybe Serejai can confirm that or deny that, or call me a retard for it because something is wrong. No hard feelings.

[loftar]

wether that is bits or bytes you didn't really make clear

Since I wrote "MB", I clearly meant bytes. ;)

[Tonkyhonk]

or call me a retard

but your recent posts sound like the lesser of the two. No hard feelings. stop provoking further.

[g1real]

wether that is bits or bytes you didn't really make clear

Since I wrote "MB", I clearly meant bytes.

I couldn't recall, so my bad then.

but your recent posts sound like the lesser of the two

Lesser of two what?

[mvgulik]

It would be nice if you could follow the history of bad tracks. But considering HDD are pretty cheap, probably not worth the effort (unless your a HDD manufacturer of course).