Search results

Did you configure spare disks in your raid10 setup? I would say that the write performance in raid10 with 10disks should be 5 times a single disk, which in your case means: 620 and you get 489

I was only able to install virtio 0.49 drivers for win 2003 r2 guests.. and they work ok ( disk virtio, net virtio )

The solution is to use a genuine installer. I was given by a customer a cracked version once ( they said by mistake ) to install it on a VM and it simply didn't work after the mangling. Once I was given the proper installer, everything was fine..

Can you issue a fio iometer using a directory on the single drive?

Re: lvremove 'vg2/vm-102-disk-1' error: Unable to deactivate logical volume "vm-102-d As the error message states , the resource is busy, it means there is something using it. Try a fuser -m /dev/vg2/vm-102-disk-1 to find out what process is using it. You might still have a running VM on it?

Is this output from the host or the container?

it's Gb/s not GB/s

It doesn't work.. the app is called RemoteViewer. I tried to launch it from terminal, I tried assign it as a default app from finder, it just launches and does nothing else, I guess it's a different app.

Of course I have the spice client.. it seems to be different and it doesn't have an option to load a script, or I couldn't find a way..

Leave "iostat 1" running for a while and see what device may cause high "tps" or just high transfer rate. Or if you have some high network throughput with processing maybe you have network I/O.. you can also check /proc/interrupts a few times in a row to see what device is triggered for interrupts..

I can use the console on osx 10.9 with safari ( you need 64-bit browser to start java, because java is only 64-bit on osx ) but I couldn't figure out how to use SPICE. On windows I just launch the spiceproxy.vv and it starts it? but on osx?

I am not sure if I understood you correctly, but if you want a pass-through drive you can just add a line in your conf with something like: ide3: /dev/sdx

The time can't have anything to do with iops performance. As mir said vmdk would give you a performance penalty.. but still I would be worried regarding the gross write iops performance..

It works nicely, thank you. I am trying different worst case scenario now to see if the data gets corrupted.. Thank you!

Can you check for what is the RAM used? maybe is caching ? then it is normal.. it would cache as much as possible, but it would free as soon as some process needs ram

Ok, but in this case you don't have an option for live migration right?

I saw this solution mentioned by udo on another topic: http://forum.proxmox.com/threads/14680-Accidentially-deleted-running-VM-with-WinSCP-how-to-rstore-image So find out the process id of kvm on the host, issue a kill -stop process, then find the file descriptor of the image by issuing: ls -l...

It hit me now when I was reconfiguring my setup to be much like yours, that you really meant LVM instead of CLVM. I am not sure if I understand this properly.. are you sure you have no problems because you are not using cluster locking? I understood the point that you have 2 separate resources...

Thinking more about your results, this cannot be a RAID10 setup, you should have pretty much the same iops read/write, this looks like a RAID-5 or 6 setup..

I'd say that the performance is just a bit over half of what it could be, are you sure there is no additional I/O when doing fio? The only thing missing in your perc conf would be some readahead policy for read, but still the write performance is only 456 iops. Could you test one disk separately...