Hey gang, thanks for all the great opinions about how I should deploy disks. Fascinating stuff. It must be really neat living in the world where things are only done The Correct Way(tm, pat pend, offer not valid in Nebraska or Louth) and you have hardware with all the gubbins you desire - and a budget to match. Seriously, great place. Must visit. In my case I have some HP Microservers with four disks each. I pair up two disks in each to make two RAID1 devices and then put LVM on top. Sometimes a disk dies. Sad but true. And sometimes I'm a bit busy so it would be nice to plug in an external USB drive and start the process of replacing the dead disk without having to power down the server and replace the disk. The microserver lacks hot-swappable SATA so USB is the only option. I do not have cleaning staff for my house - my cleaning budget mainly goes to cleaning staff for my moon base. So I can rest easy about people accidentally removing that hot-swapped USB disk and instead stress about the ridiculous health, safety and labour regulations on Luna. This bug however affects things beyond mixed SATA/USB deployments. There are discussions about it showing up with dm-crypt for instance. And yes, it is very common for people to deploy LVM on top of mdadm RAID devices. So working out such cases out might be nice. And yes, it's very possible it's not in the RAID system, but some feedback from yourselves might move solutions along. Anyway, you guys do you. I just offered some feedback where data was being silently corrupted which seems like a thing, I dunno, people doing storage software might care about. I have to get back to disk funeral planning and dealing with this complaint about bunions caused by magnetic boots. Cheers! Kevin On Sun, Jul 15, 2018 at 11:42 PM, Roger Heflin <rogerheflin@xxxxxxxxx> wrote: > Not to mention most motherboards only have at most 2 or 3 at most USB > ports. And doing high bandwidth on shared USB ports *WILL* generally > behave very unreliably. And almost all of the cards also only add a > single port (I know of one usb2.0 card that has 4 actual non-hub real > usb ports). A SATA multiplier with all of its potential issues is > a much better solution than attempting to put several disks on the > same USB hub. > > On Sun, Jul 15, 2018 at 3:23 PM, Rudy Zijlstra > <rudy@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote: >> >> >> On 2018-07-15 21:33, Kevin Lyda wrote: >>> On Sun, Jan 7, 2018 at 12:10 PM, Wols Lists <antlists@xxxxxxxxxxxxxxx> wrote: >>>> On 07/01/18 10:58, Kevin Lyda wrote: >>>>> This scenario sounds great, but last I knew mixing SATA and USB >>>>> connected disks would lead to data corruption. But I last looked >>>>> around 2010 or so. Is this still true? If not, what version of the >>>>> kernel / mdadm tools did it get fixed in. I skimmed the changelogs and >>>>> commits and didn't see any mentions that it had been. >>>> Speaking off the top of my head, I got the impression that the problem >>>> actually lies in the USB protocol. In other words it can't be fixed :-( >>> It seems it can be fixed. Are there any plans to move such >>> changes into the mainline kernel? >>> >>> >> Cannot be fixed. The big problem with USB is the not-so-dependable >> connection. This can come from multiple sources: >> - bad cable >> - bad connector (too often used for example) >> - someone who does not know what the USB is for temporarily removing it >> (for example for cleaning purposes) >> - ... >> >> Cheers >> >> Rudy >> -- >> To unsubscribe from this list: send the line "unsubscribe linux-raid" in >> the body of a message to majordomo@xxxxxxxxxxxxxxx >> More majordomo info at http://vger.kernel.org/majordomo-info.html -- Kevin Lyda Galway, Ireland -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
