You are not logged in.
Hi,
i try to explain whats my problem.
I have a Server with arch as a host. With QEMU i have a NAS (Openmediavault)
Arch, 4 Core Xeon, 32 GB Ram, 2x WD Red 4GB as Raid 0 (build in Arch)
NAS, no "static CPU, 4 GB Ram, MD0 devise, GBIT Networkadapter exclusively
Win10, 3 Cores exclusively, 14 GB exclusively, NVME as RAW device, GBIT Networkadapter exclusively
My host creates a raid 1 (MD0) which my NAS uses exclusively. So during the NAS is running i cant access (from Arch) the MD0 device. My NAS doenst see the raid members, just /dev/vdb.
Now i wanted to backup my MD0 from within the NAS. I use a 3.0 USB to SATA converter and a Seagate 8TB Archive HDD. I now have the problem that after a few 100 GB of copying data the transferrate drops. It starts with lets say 100 MB/s and after a while i drops to sometimes 1MB/s or even less.
I tried to invesigate this (attached to NAS).
Reading from the USB device like this: dd if=/dev/sda1 of=/dev/null
-> 180 MB/s
Writing to the USB device (mounted): dd if=/dev/zero of=/mnt/backup/zero3.bin bs=10M
-> 130 MB/s
This for hours
Formating the device to NTFS, 130..180 MBs (attached to NAS).
So I think the USB from within the virtual NAS isn’t my problem.
Reading from my MD0 Raid device to “null” ->140 MB/s (attached to NAS).
I think its either the NAS or my MD0 device that’s slows down. During this problems my second virtual system (Win 10) doenst have any problems.
I first recognized such problems when I tried to work a lot on my NAS (I have all my pictures on the NAS, so my editing program uses this drive). After a while the Network performance drops from 90 MB/s to maybe 5..10 MB/s. I always thought it’s a problem of my Network.
Last time my NAS crashed so badly, I wasn’t able to interrupt the copy process (no more response). So I had to kill the NAS system. Now I have “dead” files on my backup HDD which I cant delete (rm… something like “cant delete, folder not empty")
My questions is… how should I investigate this problem. In IOTOP I only see that I have a problem, but no idea where this is coming from.
Thanks
P.S. As I look into Openmediavault GUI there is a log (Performance statistics). On the cpu tab I see that the wait-io is about 100 percent. I don’t know exactly what its doing (wait-io). Does this mean my CPU waits for data that the cpu never becomes?
As im writing these lines it happens again
root@nas-vm:~# vmstat
procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu-----
r b swpd free buff cache si so bi bo in cs us sy id wa st
1 3 0 133000 1831840 3707268 0 0 33073 1452 1500 2857 4 9 56 30
A few second later it started again to copy fast
procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu-----
r b swpd free buff cache si so bi bo in cs us sy id wa st
2 0 0 117340 1840328 3722656 0 0 32713 1524 1514 2893 4 9 55 31 0
Now it stopped again (since 10 minutes) Write speed (iotop) 1..3 MB/s, vmstat:
procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu-----
r b swpd free buff cache si so bi bo in cs us sy id wa st
1 3 0 124004 1841292 3718840 0 0 31595 1559 1531 2972 4 9 53 33 0
Offline