I do this almost everyday. Do I need support on the OS? Worst case scenario, the OS crash, but your drive will be OK. Some OS might delay sending data to the drive, or saving file system information to the drive. This only applies for newly written data. Your email address will not be published. Error handling is also production-stable, but continues to be refined because by its very nature, faults are rare, therefore fault workaround and prevention testing is all the more difficult. Tangent: Host-based queueing and Native Command Queueing: Queueing is the process of sending multiple commands to a single device, without waiting for prior commands to finish.
This increases performance and reduces latency. There are three types of queueing in the ATA world:. All SATA devices are hotplug-capable. These controllers do not export enough information about the SATA phy to make it possible to support hotplug. Also, a disk powered by sata power connector that is powered with a molex to sata power adaptor can die if such molex conection is not firm or if you disconnect the molex part. The OS must know when a drive on a sata port will be disconnected prior to disconnect among modern OS are able to detect abrutly disconections and still work, other OS just freezes , so user or software apps must tell the OS that a disk is going to be removed eject on Linux, safe remove on Windows, etc.
If the OS is not designed for hot-swap, weird things can happen when you plug another disk, like the OS thinking the old disk come to live again So, yes, sata hot-swap must be implemented on sata controller hardware and on the OS To hot-swap any disk, the OS flushes all data, and send a command to the disk telling it must flush all its internal cache and then do a spin-down, after that the OS tell the sata driver to disconnect data port and if well designed also power port, then the user can safaly remove the disk no data can be sent, no power that may case electrical peaks, etc , the the sata hardware must detect infinite impedance no drive is conected on sata data port and sata power port, and get prepared for a new drive, listening for non infinite impedances, so it reconects power and data ports as soon as the other drive is connected Sata hot-swap must be implemented on hardware, oh, yes, it must be I recomend one test prior to test with the OS loaded If your bios can see the disk after hot pluging it, you know it is hot-plug If the bios detects no disk after unplug, now it is time to check another disk If such tests shows you that you hardware is not trully hot-swap, no matter what do the OS, the hot-swap procedure is very risky if not impossible at all But if your bios show you it see the drive, then it see you unpluged it, then it see the other disk plugged, then you have hardware support for hot-plug on that sata port Oh, yes, my friends, some lazy manufactures only implements hot-swap on some ports and not on all of them.
Risks of this tests you must support: Loosing sata port because if dies when pluging a drive on it while being on bios Worst case scenario: Sata root controller died because no hot-swap capable, if you where on bios, no data was on risk, but if you where running the OS you can loose all data and also damage other components like graphics card, ram, efc. Doing the tests while on bios is the safiest Said again, do not trie hot-plu, hot-remove, hot-swap, if disk is powered by the molex connector that is fixed to the disk not all drives has molex power connector, most do not have it Molex conectors are not designed for hot-plug, hot-remove, hot-swap I have bought a sata bay, which physically allows hot swapping.
So I was wondering this question. Right now I got strange results. In linux that was booted from nvme, hot swap worked. But in linux loaded from sata drive, hot swapping another drive caused reboot. I am not sure if some kernel parameters are involved, will explore later. And also will test with windows os.
However, I do not know how accurate is info from lsblk. At least, I have noticed that sdc is shown as a rotatable disk ROTA is 1 , but that is false, because it is just an sd card. I then additionally checked these things on a supermicro server without hardware raid so, just several sata ssds inserted to backplane , and they are surely hot swappable, however, RM and HOTPLUG are always 0 there too, despite in server's bios the setting for hot plug was enabled or disabled.
Sign up to join this community. The best answers are voted up and rise to the top. Stack Overflow for Teams — Collaborate and share knowledge with a private group. Use "aacraid" driver. Optional battery is available for the card's cache, for more reliable operation in the event of power loss, etc.
Card disables the drive's write cache. Supports NCQ, hot sparing. Optional battery backup unit is available. This turns out to be a Silicon Image variant or A-variant. These are rumoured to be a cheap OEM from someone else, but the chipset identity is unknown. Note Brent Norris's scathing review.
Adaptec offers proprietary drivers binary-only. See Silicon Image I so far have no information on their chipsets: This entry is a placeholder. Please see entry for Pacific Digital Talon , as no other manufacturers yet implement this open-standard hardware spec. If your desired installation kernel lacks "ahci", you may be able to use a pre-AHCI fallback mode e.
Supports the standard AHCI driver interface. Model ARC has a battery backup unit option. Erich Chen of Areca maintains GPLed source code for a "arcmsr" driver for these chipsets also available at unnet. Units use Broadcom BCM chips. The cheaper and cards differ from their and brethren in having some advanced capabilities disabled; they can be upgraded via an extra-cost software key. This entry is a placeholder.
Supported in 2. No libata driver exists for these, but Alan Cox is working on one as of Problematic proprietary Linux i binary drivers for HighPoint fakeraid release 2. Said to use the Marvell 88SX chip. HighPoint makes available proprietary driver " hptmv " source code wrapper around a binary-only proprietary core library.
Proprietary binary drivers , , , can be downloaded from the manufacturer. Correspondent Berkley Shands notes that these cards and proprietary drivers are quite CPU-intensive, even pushing a quad-Opteron system a bit. Note that models and both try to use different proprietary drivers named "hptmv6". The default read-ahead is 8 sectors. I use ; otherwise, you won't come anywhere near those numbers. An proprietary, high-performance "rrx" driver source code wrapper around binary-only proprietary core libraries, deceptively claimed to be "open source" is available from the manufacturer.
More info. Works with the kernel's gdth driver version 3. A file of hints on using the gdth driver with sundry distributions is available from ICP Vortex. One source says these cards are based on Silicon Image chips, which are probably not the system-facing chipsets , and so don't determine driver support.
See fakeraid support details under Intel ICH6.
0コメント