Either I don't understand your or it makes no sense to me. The documentation for the mentioned system tells about dual-ported NVMe drives. Unfortunately I don't see the midplane diagram there, but as I would guess dual-ported NVMe drives inserted into this system should present separate PCIe x2...
This question sounds weird to me. If you have dual port NVME then you just can connect each port to its node and use as simultaneously as file system or application running on top is able. Multipath usually means multiple connections between host and storage device. But it makes no sense for...
I'd say that your biggest problem is 4K block size. It is too small. While technically ZFS allows that, and technically there can be scenarios where it is viable, ZVOL default is 8K, and even that is found to create bottlenecks on sequential reads. Though, depending on your workload, mirrors...
I think the problem can be unrelated to send/receive itself. Tell please about configurations your old and new ZFS pools, and value of volblocksize of your zvol.
%busy in gstat spiking above 100% is often a good indicator of something being wrong (there are some exceptions, such as disk spin-up or some other legally too long operations). It means that some commands take more time then a single gstat update period, that make it jump between 0 and >100%...
It could be so if this feature would have use reference counter, as IIRC some others do. But I haven't found it while quickly grepping through the code, so I guess this may not work. Though may be I missed something.
The error message looks odd to me. Make sure that your boot partition is big enough to fit the 12'th /boot/gptzfsboot. After that check that your boot pool indeed has the mentioned file. I guess you may need that file from 12 too.
That is actually the goal of the whole process, since there...
Unfortunately disabling SHA512 checksum for dataset does not mean it is no longer used for some of already written blocks. IIRC some of ZFS features allow return from active state back to enabled when the last object of the kind is deleted, but looking on code in this case it does not look like...
Boot loader on 11-stable indeed does not support sha512 checksums, since that increases loader size a lot, that may break POLA. The main question I see here is whether you really have sha512 used on your pool. If yes, then failure is correct. Try to run `zpool get feature@sha512`.
According bhyve man page there is sectorsize=logical[/physical] option, which should allow overriding physical sector to 4KB without causing issues for pool operation.
Yes, I've heard that MS SQL does not like disks with physical sectors other then 512b and 4KB. Your solution has two downsides: 1) small block size increase data fragmentation and reduces performance on linear read operations, and 2) with some pool configurations it may lead to very bad space...
Windows 7 (same as all desktop OS from Microsoft) does not support SCSI multi-pathing. Only Windows Server versions (IIRC starting from 2003) support it.
Just a guess, depending on what is your storage, restore from backup may defragment the data on a storage, that may reduce I/O latencies to reasonable level.
According to man page, snd_emu10kx driver has two modes, exposing channel pairs as separate devices or as one multichannel device. It was done in times when FreeBSD had no multichannel audio support. You may try to switch the driver into multichannel mode.
Also don't forget to switch virtual...
First check that FreeBSD received resize event for the disk checking geom disk list da0. If yes, then use gpart resize -i 3 da0. After that read about ZFS autoexpand.
Deeper states can be blocked by selection of active timer. LAPIC timer has problems with C3 states on old systems, so those states may be disabled when it is active. You may try to force different timer hardware to use with kern.eventtimer.timer sysctl if you really want deeper C states to be...
Combination of eSATA controller and external enclosure is pretty cheap, but that is the only benefit of it. While FreeBSD supports that kind of hardware, due to its low cost it is also often known by its low reliability. Performance may also suffer, since IIRC this ASMedia chip does not...
I'm afraid with different number of channels at least minimal diver modification may be required, since the driver is very small and so probably very dumb.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.