On 01/26/2014 02:37 AM, Steve Dainard wrote:

Thanks for the responses everyone, really appreciate it.

I've condensed the other questions into this reply.

Steve, What is the CPU load of the GlusterFS host when comparing the raw brick test to the gluster mount point test? Give it 30 seconds and see what top reports. You’ll probably have to significantly increase the count on the test so that it runs that long.

- Nick

Gluster mount point:

*4K* on GLUSTER host [root@gluster1 rep2]# dd if=/dev/zero of=/mnt/rep2/test1 bs=4k count=500000 500000+0 records in 500000+0 records out 2048000000 <tel:2048000000> bytes (2.0 GB) copied, 100.076 s, 20.5 MB/s

Top reported this right away: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 1826 root 20 0 294m 33m 2540 S 27.2 0.4 0:04.31 glusterfs 2126 root 20 0 1391m 31m 2336 S 22.6 0.4 11:25.48 glusterfsd

Then at about 20+ seconds top reports this: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 1826 root 20 0 294m 35m 2660 R 141.7 0.5 1:14.94 glusterfs 2126 root 20 0 1392m 31m 2344 S 33.7 0.4 11:46.56 glusterfsd

*4K* Directly on the brick: dd if=/dev/zero of=test1 bs=4k count=500000 500000+0 records in 500000+0 records out 2048000000 <tel:2048000000> bytes (2.0 GB) copied, 4.99367 s, 410 MB/s

7750 root 20 0 102m 648 544 R 50.3 0.0 0:01.52 dd 7719 root 20 0 0 0 0 D 1.0 0.0 0:01.50 flush-253:2

Same test, gluster mount point on OVIRT host: dd if=/dev/zero of=/mnt/rep2/test1 bs=4k count=500000 500000+0 records in 500000+0 records out 2048000000 <tel:2048000000> bytes (2.0 GB) copied, 42.4518 s, 48.2 MB/s

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 2126 root 20 0 1396m 31m 2360 S 40.5 0.4 13:28.89 glusterfsd

Same test, on OVIRT host but against NFS mount point: dd if=/dev/zero of=/mnt/rep2-nfs/test1 bs=4k count=500000 500000+0 records in 500000+0 records out 2048000000 <tel:2048000000> bytes (2.0 GB) copied, 18.8911 s, 108 MB/s

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 2141 root 20 0 550m 184m 2840 R 84.6 2.3 16:43.10 glusterfs 2126 root 20 0 1407m 30m 2368 S 49.8 0.4 13:49.07 glusterfsd

Interesting - It looks like if I use a NFS mount point, I incur a cpu hit on two processes instead of just the daemon. I also get much better performance if I'm not running dd (fuse) on the GLUSTER host.

The storage servers are a bit older, but are both dual socket quad core

opterons with 4x 7200rpm drives.

A block size of 4k is quite small so that the context switch overhead involved with fuse would be more perceivable.

Would it be possible to increase the block size for dd and test?

I'm in the process of setting up a share from my desktop and I'll see if

I can bench between the two systems. Not sure if my ssd will impact the

tests, I've heard there isn't an advantage using ssd storage for glusterfs.

Do you have any pointers to this source of information? Typically glusterfs performance for virtualization work loads is bound by the slowest element in the entire stack. Usually storage/disks happen to be the bottleneck and ssd storage does benefit glusterfs.

-Vijay

I had a couple technical calls with RH (re: RHSS), and when I asked if SSD's could add any benefit I was told no. The context may have been in a product comparison to other storage vendors, where they use SSD's for read/write caching, versus having an all SSD storage domain (which I'm not proposing, but which is effectively what my desktop would provide).

Increasing bs against NFS mount point (gluster backend): dd if=/dev/zero of=/mnt/rep2-nfs/test1 bs=128k count=16000 16000+0 records in 16000+0 records out 2097152000 <tel:2097152000> bytes (2.1 GB) copied, 19.1089 s, 110 MB/s

GLUSTER host top reports: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 2141 root 20 0 550m 183m 2844 R 88.9 2.3 17:30.82 glusterfs 2126 root 20 0 1414m 31m 2408 S 46.1 0.4 14:18.18 glusterfsd

So roughly the same performance as 4k writes remotely. I'm guessing if I could randomize these writes we'd see a large difference.

Check this thread out, http://raobharata.wordpress.com/2012/10/29/qemu-glusterfs-native-integration... it's quite dated but I remember seeing similar figures.

In fact when I used FIO on a libgfapi mounted VM I got slightly faster read/write speeds than on the physical box itself (I assume because of some level of caching). On NFS it was close to half.. You'll probably get a little more interesting results using FIO opposed to dd

( -Andrew)

Sorry Andrew, I meant to reply to your other message - it looks like CentOS 6.5 can't use libgfapi right now, I stumbled across this info in a couple threads. Something about how the CentOS build has different flags set on build for RHEV snapshot support then RHEL, so native gluster storage domains are disabled because snapshot support is assumed and would break otherwise. I'm assuming this is still valid as I cannot get a storage lock when I attempt a gluster storage domain.

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

I've setup a NFS storage domain on my desktops SSD. I've re-installed win 2008 r2 and initially it was running smoother.

Disk performance peaks at 100MB/s.

If I copy a 250MB file from a share into the Windows VM, it writes out quickly, less than 5 seconds.

If I copy 20 files, ranging in file sizes from 4k to 200MB, totaling in 650MB from the share - windows becomes unresponsive, in top the desktop's nfs daemon is barely being touched at all, and then eventually is not hit. I can still interact with the VM's windows through the spice console. Eventually the file transfer will start and rocket through the transfer.

I've opened a 271MB zip file with 4454 files and started the extract process but the progress windows will sit on 'calculating...' after a significant period of time the decompression starts and runs at <200KB/second. Windows is guesstimating 1HR completion time. Eventually even this freezes up, and my spice console mouse won't grab. I can still see the resource monitor in the Windows VM doing its thing but have to poweroff the VM as its no longer usable.

The windows update process is the same. It seems like when the guest needs quick large writes its fine, but lots of io causes serious hanging, unresponsiveness, spice mouse cursor freeze, and eventually poweroff/reboot is the only way to get it back.

Also, during window 2008 r2 install the 'expanding windows files' task is quite slow, roughly 1% progress every 20 seconds (~30 mins to complete). The GLUSTER host shows these stats pretty consistently: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 8139 root 20 0 1380m 28m 2476 R 83.1 0.4 8:35.78 glusterfsd 8295 root 20 0 550m 186m 2980 S 4.3 2.4 1:52.56 glusterfs

bwm-ng v0.6 (probing every 2.000s), press 'h' for help input: /proc/net/dev type: rate \ iface Rx Tx Total

============================================================================== lo: 3719.31 KB/s 3719.31 KB/s 7438.62 KB/s eth0: 3405.12 KB/s 3903.28 KB/s 7308.40 KB/s

I've copied the same zip file to an nfs mount point on the OVIRT host (gluster backend) and get about 25 - 600 KB/s during unzip. The same test on NFS mount point (desktop SSD ext4 backend) averaged a network transfer speed of 5MB/s and completed in about 40 seconds.

I have a RHEL 6.5 guest running on the NFS/gluster backend storage domain, and just did the same test. Extracting the file took 22.3 seconds (faster than the fuse mount point on the host !?!?).

GLUSTER host top reported this while the RHEL guest was decompressing the zip file: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 2141 root 20 0 555m 187m 2844 S 4.0 2.4 18:17.00 glusterfs 2122 root 20 0 1380m 31m 2396 S 2.3 0.4 83:19.40 glusterfsd

*Steve Dainard * IT Infrastructure Manager Miovision <http://miovision.com/> | /Rethink Traffic/ 519-513-2407 <tel:519-513-2407> ex.250 877-646-8476 <tel:877-646-8476> (toll-free)

*Blog <http://miovision.com/blog> | **LinkedIn <https://www.linkedin.com/company/miovision-technologies> | Twitter <https://twitter.com/miovision> | Facebook <https://www.facebook.com/miovision>* ------------------------------------------------------------------------ Miovision Technologies Inc. | 148 Manitou Drive, Suite 101, Kitchener, ON, Canada | N2C 1L3 This e-mail may contain information that is privileged or confidential. If you are not the intended recipient, please delete the e-mail and any attachments and notify us immediately.

_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users

please note currently (>3.3.1), we don't use libgfapi on fedora as well, as we found some gaps in functionality in the libvirt libgfapi support for snapshots. once these are resolved, we can re-enable libgfapi on a glusterfs storage domain.

This is a multi-part message in MIME format. --------------080505030003020501080002 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Adding the virtio-scsi developers. Anyhow, virtio-scsi is newer and less established than viostor (the block device), so you might want to try it out. A disclaimer: There are time and patches gaps between RHEL and other versions. Ronen. On 01/28/2014 10:39 PM, Steve Dainard wrote:

I've had a bit of luck here.

Overall IO performance is very poor during Windows updates, but a contributing factor seems to be the "SCSI Controller" device in the guest. This last install I didn't install a driver for that device, and my performance is much better. Updates still chug along quite slowly, but I seem to have more than the < 100KB/s write speeds I was seeing previously.

Does anyone know what this device is for? I have the "Red Hat VirtIO SCSI Controller" listed under storage controllers.

*Steve Dainard * IT Infrastructure Manager Miovision <http://miovision.com/> | /Rethink Traffic/ 519-513-2407 ex.250 877-646-8476 (toll-free)

*Blog <http://miovision.com/blog> | **LinkedIn <https://www.linkedin.com/company/miovision-technologies> | Twitter <https://twitter.com/miovision> | Facebook <https://www.facebook.com/miovision>* ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ Miovision Technologies Inc. | 148 Manitou Drive, Suite 101, Kitchener, ON, Canada | N2C 1L3 This e-mail may contain information that is privileged or confidential. If you are not the intended recipient, please delete the e-mail and any attachments and notify us immediately.

On Sun, Jan 26, 2014 at 2:33 AM, Itamar Heim <iheim@redhat.com <mailto:iheim@redhat.com>> wrote:

On 01/26/2014 02:37 AM, Steve Dainard wrote:

Thanks for the responses everyone, really appreciate it.

I've condensed the other questions into this reply.

Steve, What is the CPU load of the GlusterFS host when comparing the raw brick test to the gluster mount point test? Give it 30 seconds and see what top reports. You'll probably have to significantly increase the count on the test so that it runs that long.

- Nick

Gluster mount point:

*4K* on GLUSTER host [root@gluster1 rep2]# dd if=/dev/zero of=/mnt/rep2/test1 bs=4k count=500000 500000+0 records in 500000+0 records out 2048000000 <tel:2048000000> <tel:2048000000 <tel:2048000000>> bytes (2.0 GB) copied, 100.076 s, 20.5 MB/s

Top reported this right away: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 1826 root 20 0 294m 33m 2540 S 27.2 0.4 0:04.31 glusterfs 2126 root 20 0 1391m 31m 2336 S 22.6 0.4 11:25.48 glusterfsd

Then at about 20+ seconds top reports this: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 1826 root 20 0 294m 35m 2660 R 141.7 0.5 1:14.94 glusterfs 2126 root 20 0 1392m 31m 2344 S 33.7 0.4 11:46.56 glusterfsd

*4K* Directly on the brick: dd if=/dev/zero of=test1 bs=4k count=500000 500000+0 records in 500000+0 records out 2048000000 <tel:2048000000> <tel:2048000000 <tel:2048000000>> bytes (2.0 GB) copied, 4.99367 s, 410 MB/s

7750 root 20 0 102m 648 544 R 50.3 0.0 0:01.52 dd 7719 root 20 0 0 0 0 D 1.0 0.0 0:01.50 flush-253:2

Same test, gluster mount point on OVIRT host: dd if=/dev/zero of=/mnt/rep2/test1 bs=4k count=500000 500000+0 records in 500000+0 records out 2048000000 <tel:2048000000> <tel:2048000000 <tel:2048000000>> bytes (2.0 GB) copied, 42.4518 s, 48.2 MB/s

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 2126 root 20 0 1396m 31m 2360 S 40.5 0.4 13:28.89 glusterfsd

Same test, on OVIRT host but against NFS mount point: dd if=/dev/zero of=/mnt/rep2-nfs/test1 bs=4k count=500000 500000+0 records in 500000+0 records out 2048000000 <tel:2048000000> <tel:2048000000 <tel:2048000000>> bytes (2.0 GB) copied, 18.8911 s, 108 MB/s

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 2141 root 20 0 550m 184m 2840 R 84.6 2.3 16:43.10 glusterfs 2126 root 20 0 1407m 30m 2368 S 49.8 0.4 13:49.07 glusterfsd

Interesting - It looks like if I use a NFS mount point, I incur a cpu hit on two processes instead of just the daemon. I also get much better performance if I'm not running dd (fuse) on the GLUSTER host.

The storage servers are a bit older, but are both dual socket quad core

opterons with 4x 7200rpm drives.

A block size of 4k is quite small so that the context switch overhead involved with fuse would be more perceivable.

Would it be possible to increase the block size for dd and test?

I'm in the process of setting up a share from my desktop and I'll see if

I can bench between the two systems. Not sure if my ssd will impact the

tests, I've heard there isn't an advantage using ssd storage for glusterfs.

Do you have any pointers to this source of information? Typically glusterfs performance for virtualization work loads is bound by the slowest element in the entire stack. Usually storage/disks happen to be the bottleneck and ssd storage does benefit glusterfs.

-Vijay

I had a couple technical calls with RH (re: RHSS), and when I asked if SSD's could add any benefit I was told no. The context may have been in a product comparison to other storage vendors, where they use SSD's for read/write caching, versus having an all SSD storage domain (which I'm not proposing, but which is effectively what my desktop would provide).

Increasing bs against NFS mount point (gluster backend): dd if=/dev/zero of=/mnt/rep2-nfs/test1 bs=128k count=16000 16000+0 records in 16000+0 records out 2097152000 <tel:2097152000> <tel:2097152000 <tel:2097152000>> bytes (2.1 GB) copied, 19.1089 s, 110 MB/s

GLUSTER host top reports: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 2141 root 20 0 550m 183m 2844 R 88.9 2.3 17:30.82 glusterfs 2126 root 20 0 1414m 31m 2408 S 46.1 0.4 14:18.18 glusterfsd

So roughly the same performance as 4k writes remotely. I'm guessing if I could randomize these writes we'd see a large difference.

Check this thread out, http://raobharata.wordpress.com/2012/10/29/qemu-glusterfs-native-integration... it's quite dated but I remember seeing similar figures.

In fact when I used FIO on a libgfapi mounted VM I got slightly faster read/write speeds than on the physical box itself (I assume because of some level of caching). On NFS it was close to half.. You'll probably get a little more interesting results using FIO opposed to dd

( -Andrew)

Sorry Andrew, I meant to reply to your other message - it looks like CentOS 6.5 can't use libgfapi right now, I stumbled across this info in a couple threads. Something about how the CentOS build has different flags set on build for RHEV snapshot support then RHEL, so native gluster storage domains are disabled because snapshot support is assumed and would break otherwise. I'm assuming this is still valid as I cannot get a storage lock when I attempt a gluster storage domain.

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

I've setup a NFS storage domain on my desktops SSD. I've re-installed win 2008 r2 and initially it was running smoother.

Disk performance peaks at 100MB/s.

If I copy a 250MB file from a share into the Windows VM, it writes out quickly, less than 5 seconds.

If I copy 20 files, ranging in file sizes from 4k to 200MB, totaling in 650MB from the share - windows becomes unresponsive, in top the desktop's nfs daemon is barely being touched at all, and then eventually is not hit. I can still interact with the VM's windows through the spice console. Eventually the file transfer will start and rocket through the transfer.

I've opened a 271MB zip file with 4454 files and started the extract process but the progress windows will sit on 'calculating...' after a significant period of time the decompression starts and runs at <200KB/second. Windows is guesstimating 1HR completion time. Eventually even this freezes up, and my spice console mouse won't grab. I can still see the resource monitor in the Windows VM doing its thing but have to poweroff the VM as its no longer usable.

The windows update process is the same. It seems like when the guest needs quick large writes its fine, but lots of io causes serious hanging, unresponsiveness, spice mouse cursor freeze, and eventually poweroff/reboot is the only way to get it back.

Also, during window 2008 r2 install the 'expanding windows files' task is quite slow, roughly 1% progress every 20 seconds (~30 mins to complete). The GLUSTER host shows these stats pretty consistently: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 8139 root 20 0 1380m 28m 2476 R 83.1 0.4 8:35.78 glusterfsd 8295 root 20 0 550m 186m 2980 S 4.3 2.4 1:52.56 glusterfs

bwm-ng v0.6 (probing every 2.000s), press 'h' for help input: /proc/net/dev type: rate \ iface Rx Tx Total

============================================================================== lo: 3719.31 KB/s 3719.31 KB/s 7438.62 KB/s eth0: 3405.12 KB/s 3903.28 KB/s 7308.40 KB/s

I've copied the same zip file to an nfs mount point on the OVIRT host (gluster backend) and get about 25 - 600 KB/s during unzip. The same test on NFS mount point (desktop SSD ext4 backend) averaged a network transfer speed of 5MB/s and completed in about 40 seconds.

I have a RHEL 6.5 guest running on the NFS/gluster backend storage domain, and just did the same test. Extracting the file took 22.3 seconds (faster than the fuse mount point on the host !?!?).

GLUSTER host top reported this while the RHEL guest was decompressing the zip file: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 2141 root 20 0 555m 187m 2844 S 4.0 2.4 18:17.00 glusterfs 2122 root 20 0 1380m 31m 2396 S 2.3 0.4 83:19.40 glusterfsd

*Steve Dainard * IT Infrastructure Manager Miovision <http://miovision.com/> | /Rethink Traffic/ 519-513-2407 <tel:519-513-2407> <tel:519-513-2407 <tel:519-513-2407>> ex.250 877-646-8476 <tel:877-646-8476> <tel:877-646-8476 <tel:877-646-8476>> (toll-free)

*Blog <http://miovision.com/blog> | **LinkedIn <https://www.linkedin.com/company/miovision-technologies> | Twitter <https://twitter.com/miovision> | Facebook <https://www.facebook.com/miovision>* ------------------------------------------------------------------------ Miovision Technologies Inc. | 148 Manitou Drive, Suite 101, Kitchener, ON, Canada | N2C 1L3 This e-mail may contain information that is privileged or confidential. If you are not the intended recipient, please delete the e-mail and any attachments and notify us immediately.

_______________________________________________ Users mailing list Users@ovirt.org <mailto:Users@ovirt.org> http://lists.ovirt.org/mailman/listinfo/users

please note currently (>3.3.1), we don't use libgfapi on fedora as well, as we found some gaps in functionality in the libvirt libgfapi support for snapshots. once these are resolved, we can re-enable libgfapi on a glusterfs storage domain.

--------------080505030003020501080002 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit <html> <head> <meta content="text/html; charset=ISO-8859-1" http-equiv="Content-Type"> </head> <body bgcolor="#FFFFFF" text="#000000"> <div class="moz-cite-prefix">Adding the virtio-scsi developers.<br> Anyhow, virtio-scsi is newer and less established than viostor (the block device), so you might want to try it out.<br> A disclaimer: There are time and patches gaps between RHEL and other versions.<br> <br> Ronen.<br> <br> On 01/28/2014 10:39 PM, Steve Dainard wrote:<br> </div> <blockquote cite="mid:CAHnsdUt1wPNcWFhy1eW8iYRrp4cNn=9WMFt8sVV5bYGJS1cvUw@mail.gmail.com" type="cite"> <div dir="ltr">I've had a bit of luck here. <div><br> </div> <div>Overall IO performance is very poor during Windows updates, but a contributing factor seems to be the "SCSI Controller" device in the guest. This last install I didn't install a driver for that device, and my performance is much better. Updates still chug along quite slowly, but I seem to have more than the < 100KB/s write speeds I was seeing previously.</div> <div><br> </div> <div>Does anyone know what this device is for? I have the "Red Hat VirtIO SCSI Controller" listed under storage controllers.</div> </div> <div class="gmail_extra"><br clear="all"> <div> <div dir="ltr"> <span style="font-family:arial,sans-serif;font-size:16px"><strong>Steve Dainard </strong></span><span style="font-size:12px"></span><br> <span style="font-family:arial,sans-serif;font-size:12px">IT Infrastructure Manager<br> <a moz-do-not-send="true" href="http://miovision.com/" target="_blank">Miovision</a> | <em>Rethink Traffic</em><br> 519-513-2407 ex.250<br> 877-646-8476 (toll-free)<br> <br> <strong style="font-family:arial,sans-serif;font-size:13px;color:#999999"><a moz-do-not-send="true" href="http://miovision.com/blog" target="_blank">Blog</a>  |  </strong><font style="font-family:arial,sans-serif;font-size:13px" color="#999999"><strong><a moz-do-not-send="true" href="https://www.linkedin.com/company/miovision-technologies" target="_blank">LinkedIn</a>  |  <a moz-do-not-send="true" href="https://twitter.com/miovision" target="_blank">Twitter</a>  |  <a moz-do-not-send="true" href="https://www.facebook.com/miovision" target="_blank">Facebook</a></strong></font> </span> <hr style="font-family:arial,sans-serif;font-size:13px;color:#333333;clear:both"> <div style="color:#999999;font-family:arial,sans-serif;font-size:13px;padding-top:5px"> <span style="font-family:arial,sans-serif;font-size:12px">Miovision Technologies Inc. | 148 Manitou Drive, Suite 101, Kitchener, ON, Canada | N2C 1L3</span><br> <span style="font-family:arial,sans-serif;font-size:12px">This e-mail may contain information that is privileged or confidential. If you are not the intended recipient, please delete the e-mail and any attachments and notify us immediately.</span></div> </div> </div> <br> <br> <div class="gmail_quote">On Sun, Jan 26, 2014 at 2:33 AM, Itamar Heim <span dir="ltr"><<a moz-do-not-send="true" href="mailto:iheim@redhat.com" target="_blank">iheim@redhat.com</a>></span> wrote:<br> <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> <div class="im">On 01/26/2014 02:37 AM, Steve Dainard wrote:<br> </div> <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> <div class="im"> Thanks for the responses everyone, really appreciate it.<br> <br> I've condensed the other questions into this reply.<br> <br> <br>     Steve,<br>     What is the CPU load of the GlusterFS host when comparing the raw<br>     brick test to the gluster mount point test? Give it 30 seconds and<br>     see what top reports. You’ll probably have to significantly increase<br>     the count on the test so that it runs that long.<br> <br>     - Nick<br> <br> <br> <br> Gluster mount point:<br> <br> *4K* on GLUSTER host<br> [root@gluster1 rep2]# dd if=/dev/zero of=/mnt/rep2/test1 bs=4k count=500000<br> 500000+0 records in<br> 500000+0 records out<br> </div> <a moz-do-not-send="true" href="tel:2048000000" value="+12048000000" target="_blank">2048000000</a> <tel:<a moz-do-not-send="true" href="tel:2048000000" value="+12048000000" target="_blank">2048000000</a>> bytes (2.0 GB) copied, 100.076 s, 20.5 MB/s <div class="im"> <br> <br> Top reported this right away:<br> PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND<br>   1826 root      20   0  294m  33m 2540 S 27.2  0.4   0:04.31 glusterfs<br>   2126 root      20   0 1391m  31m 2336 S 22.6  0.4  11:25.48 glusterfsd<br> <br> Then at about 20+ seconds top reports this:<br>    PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND<br>   1826 root      20   0  294m  35m 2660 R 141.7  0.5   1:14.94 glusterfs<br>   2126 root      20   0 1392m  31m 2344 S 33.7  0.4  11:46.56 glusterfsd<br> <br> *4K* Directly on the brick:<br> dd if=/dev/zero of=test1 bs=4k count=500000<br> 500000+0 records in<br> 500000+0 records out<br> </div> <a moz-do-not-send="true" href="tel:2048000000" value="+12048000000" target="_blank">2048000000</a> <tel:<a moz-do-not-send="true" href="tel:2048000000" value="+12048000000" target="_blank">2048000000</a>> bytes (2.0 GB) copied, 4.99367 s, 410 MB/s <div class="im"> <br> <br>   7750 root      20   0  102m  648  544 R 50.3  0.0   0:01.52 dd<br>   7719 root      20   0     0    0    0 D  1.0  0.0   0:01.50 flush-253:2<br> <br> Same test, gluster mount point on OVIRT host:<br> dd if=/dev/zero of=/mnt/rep2/test1 bs=4k count=500000<br> 500000+0 records in<br> 500000+0 records out<br> </div> <a moz-do-not-send="true" href="tel:2048000000" value="+12048000000" target="_blank">2048000000</a> <tel:<a moz-do-not-send="true" href="tel:2048000000" value="+12048000000" target="_blank">2048000000</a>> bytes (2.0 GB) copied, 42.4518 s, 48.2 MB/s <div class="im"> <br> <br>    PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND<br>   2126 root      20   0 1396m  31m 2360 S 40.5  0.4  13:28.89 glusterfsd<br> <br> <br> Same test, on OVIRT host but against NFS mount point:<br> dd if=/dev/zero of=/mnt/rep2-nfs/test1 bs=4k count=500000<br> 500000+0 records in<br> 500000+0 records out<br> </div> <a moz-do-not-send="true" href="tel:2048000000" value="+12048000000" target="_blank">2048000000</a> <tel:<a moz-do-not-send="true" href="tel:2048000000" value="+12048000000" target="_blank">2048000000</a>> bytes (2.0 GB) copied, 18.8911 s, 108 MB/s <div> <div class="h5"> <br> <br> PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND<br>   2141 root      20   0  550m 184m 2840 R 84.6  2.3  16:43.10 glusterfs<br>   2126 root      20   0 1407m  30m 2368 S 49.8  0.4  13:49.07 glusterfsd<br> <br> Interesting - It looks like if I use a NFS mount point, I incur a cpu<br> hit on two processes instead of just the daemon. I also get much better<br> performance if I'm not running dd (fuse) on the GLUSTER host.<br> <br> <br>         The storage servers are a bit older, but are both dual socket<br>         quad core<br> <br>         opterons with 4x 7200rpm drives.<br> <br> <br>     A block size of 4k is quite small so that the context switch<br>     overhead involved with fuse would be more perceivable.<br> <br>     Would it be possible to increase the block size for dd and test?<br> <br> <br> <br>         I'm in the process of setting up a share from my desktop and<br>         I'll see if<br> <br>         I can bench between the two systems. Not sure if my ssd will<br>         impact the<br> <br>         tests, I've heard there isn't an advantage using ssd storage for<br>         glusterfs.<br> <br> <br>     Do you have any pointers to this source of information? Typically<br>     glusterfs performance for virtualization work loads is bound by the<br>     slowest element in the entire stack. Usually storage/disks happen to<br>     be the bottleneck and ssd storage does benefit glusterfs.<br> <br>     -Vijay<br> <br> <br> I had a couple technical calls with RH (re: RHSS), and when I asked if<br> SSD's could add any benefit I was told no. The context may have been in<br> a product comparison to other storage vendors, where they use SSD's for<br> read/write caching, versus having an all SSD storage domain (which I'm<br> not proposing, but which is effectively what my desktop would provide).<br> <br> Increasing bs against NFS mount point (gluster backend):<br> dd if=/dev/zero of=/mnt/rep2-nfs/test1 bs=128k count=16000<br> 16000+0 records in<br> 16000+0 records out<br> </div> </div> <a moz-do-not-send="true" href="tel:2097152000" value="+12097152000" target="_blank">2097152000</a> <tel:<a moz-do-not-send="true" href="tel:2097152000" value="+12097152000" target="_blank">2097152000</a>> bytes (2.1 GB) copied, 19.1089 s, 110 MB/s <div> <div class="h5"><br> <br> <br> GLUSTER host top reports:<br>    PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND<br>   2141 root      20   0  550m 183m 2844 R 88.9  2.3  17:30.82 glusterfs<br>   2126 root      20   0 1414m  31m 2408 S 46.1  0.4  14:18.18 glusterfsd<br> <br> So roughly the same performance as 4k writes remotely. I'm guessing if I<br> could randomize these writes we'd see a large difference.<br> <br> <br>     Check this thread out,<br>     <a moz-do-not-send="true" href="http://raobharata.wordpress.com/2012/10/29/qemu-glusterfs-native-integration..." target="_blank">http://raobharata.wordpress.com/2012/10/29/qemu-glusterfs-native-integration/</a> it's<br>     quite dated but I remember seeing similar figures.<br> <br>     In fact when I used FIO on a libgfapi mounted VM I got slightly<br>     faster read/write speeds than on the physical box itself (I assume<br>     because of some level of caching). On NFS it was close to half..<br>     You'll probably get a little more interesting results using FIO<br>     opposed to dd<br> <br>     ( -Andrew)<br> <br> <br> Sorry Andrew, I meant to reply to your other message - it looks like<br> CentOS 6.5 can't use libgfapi right now, I stumbled across this info in<br> a couple threads. Something about how the CentOS build has different<br> flags set on build for RHEV snapshot support then RHEL, so native<br> gluster storage domains are disabled because snapshot support is assumed<br> and would break otherwise. I'm assuming this is still valid as I cannot<br> get a storage lock when I attempt a gluster storage domain.<br> <br> <br> --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------<br> <br> I've setup a NFS storage domain on my desktops SSD. I've re-installed<br> win 2008 r2 and initially it was running smoother.<br> <br> Disk performance peaks at 100MB/s.<br> <br> If I copy a 250MB file from a share into the Windows VM, it writes out<br> quickly, less than 5 seconds.<br> <br> If I copy 20 files, ranging in file sizes from 4k to 200MB, totaling in<br> 650MB from the share - windows becomes unresponsive, in top the<br> desktop's nfs daemon is barely being touched at all, and then eventually<br> is not hit. I can still interact with the VM's windows through the spice<br> console. Eventually the file transfer will start and rocket through the<br> transfer.<br> <br> I've opened a 271MB zip file with 4454 files and started the extract<br> process but the progress windows will sit on 'calculating...' after a<br> significant period of time the decompression starts and runs at<br> <200KB/second. Windows is guesstimating 1HR completion time. Eventually<br> even this freezes up, and my spice console mouse won't grab. I can still<br> see the resource monitor in the Windows VM doing its thing but have to<br> poweroff the VM as its no longer usable.<br> <br> The windows update process is the same. It seems like when the guest<br> needs quick large writes its fine, but lots of io causes serious<br> hanging, unresponsiveness, spice mouse cursor freeze, and eventually<br> poweroff/reboot is the only way to get it back.<br> <br> Also, during window 2008 r2 install the 'expanding windows files' task<br> is quite slow, roughly 1% progress every 20 seconds (~30 mins to<br> complete). The GLUSTER host shows these stats pretty consistently:<br> PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND<br>   8139 root      20   0 1380m  28m 2476 R 83.1  0.4   8:35.78 glusterfsd<br>   8295 root      20   0  550m 186m 2980 S  4.3  2.4   1:52.56 glusterfs<br> <br> bwm-ng v0.6 (probing every 2.000s), press 'h' for help<br> input: /proc/net/dev type: rate<br>    \         iface                   Rx                   Tx<br>     Total<br> <br> ==============================================================================<br>                 lo:        3719.31 KB/s         3719.31 KB/s<br> 7438.62 KB/s<br>               eth0:        3405.12 KB/s         3903.28 KB/s<br> 7308.40 KB/s<br> <br> <br> I've copied the same zip file to an nfs mount point on the OVIRT host<br> (gluster backend) and get about 25 - 600 KB/s during unzip. The same<br> test on NFS mount point (desktop SSD ext4 backend) averaged a network<br> transfer speed of 5MB/s and completed in about 40 seconds.<br> <br> I have a RHEL 6.5 guest running on the NFS/gluster backend storage<br> domain, and just did the same test. Extracting the file took 22.3<br> seconds (faster than the fuse mount point on the host !?!?).<br> <br> GLUSTER host top reported this while the RHEL guest was decompressing<br> the zip file:<br>    PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND<br>   2141 root      20   0  555m 187m 2844 S  4.0  2.4  18:17.00 glusterfs<br>   2122 root      20   0 1380m  31m 2396 S  2.3  0.4  83:19.40 glusterfsd<br> <br> <br> <br> <br> <br> </div> </div> <div class="im"> *Steve Dainard *<br> IT Infrastructure Manager<br> Miovision <<a moz-do-not-send="true" href="http://miovision.com/" target="_blank">http://miovision.com/</a>>; | /Rethink Traffic/<br> <a moz-do-not-send="true" href="tel:519-513-2407" value="+15195132407" target="_blank">519-513-2407</a> <tel:<a moz-do-not-send="true" href="tel:519-513-2407" value="+15195132407" target="_blank">519-513-2407</a>> ex.250<br> <a moz-do-not-send="true" href="tel:877-646-8476" value="+18776468476" target="_blank">877-646-8476</a> <tel:<a moz-do-not-send="true" href="tel:877-646-8476" value="+18776468476" target="_blank">877-646-8476</a>> (toll-free)<br> <br> *Blog <<a moz-do-not-send="true" href="http://miovision.com/blog" target="_blank">http://miovision.com/blog</a>>;  | **LinkedIn<br> <<a moz-do-not-send="true" href="https://www.linkedin.com/company/miovision-technologies" target="_blank">https://www.linkedin.com/company/miovision-technologies</a>>;  | Twitter<br> <<a moz-do-not-send="true" href="https://twitter.com/miovision" target="_blank">https://twitter.com/miovision</a>>;  | Facebook<br> <<a moz-do-not-send="true" href="https://www.facebook.com/miovision" target="_blank">https://www.facebook.com/miovision</a>>*<br> ------------------------------------------------------------------------<br> </div> <div class="im"> Miovision Technologies Inc. | 148 Manitou Drive, Suite 101, Kitchener,<br> ON, Canada | N2C 1L3<br> This e-mail may contain information that is privileged or confidential.<br> If you are not the intended recipient, please delete the e-mail and any<br> attachments and notify us immediately.<br> <br> <br> <br> <br> </div> <div class="im"> _______________________________________________<br> Users mailing list<br> <a moz-do-not-send="true" href="mailto:Users@ovirt.org" target="_blank">Users@ovirt.org</a><br> <a moz-do-not-send="true" href="http://lists.ovirt.org/mailman/listinfo/users" target="_blank">http://lists.ovirt.org/mailman/listinfo/users</a><br> <br> </div> </blockquote> <br> please note currently (>3.3.1), we don't use libgfapi on fedora as well, as we found some gaps in functionality in the libvirt libgfapi support for snapshots. once these are resolved, we can re-enable libgfapi on a glusterfs storage domain.<br> </blockquote> </div> <br> </div> </blockquote> <br> </body> </html> --------------080505030003020501080002--

On Wed, Jan 29, 2014 at 5:11 AM, Vadim Rozenfeld <vrozenfe@redhat.com>wrote:

On Wed, 2014-01-29 at 11:30 +0200, Ronen Hod wrote:

...

Adding the virtio-scsi developers. Anyhow, virtio-scsi is newer and less established than viostor (the block device), so you might want to try it out.

[VR] Was it "SCSI Controller" or "SCSI pass-through controller"? If it's "SCSI Controller" then it will be viostor (virtio-blk) device driver.

"SCSI Controller" is listed in device manager. Hardware ID's: PCI\VEN_1AF4&DEV_1004&SUBSYS_00081AF4&REV_00 PCI\VEN_1AF4&DEV_1004&SUBSYS_00081AF4 PCI\VEN_1AF4&DEV_1004&CC_010000 PCI\VEN_1AF4&DEV_1004&CC_0100

...

A disclaimer: There are time and patches gaps between RHEL and other versions.

Ronen.

On 01/28/2014 10:39 PM, Steve Dainard wrote:

...

I've had a bit of luck here.

Overall IO performance is very poor during Windows updates, but a contributing factor seems to be the "SCSI Controller" device in the guest. This last install I didn't install a driver for that device,

[VR] Does it mean that your system disk is IDE and the data disk (virtio-blk) is not accessible?

In Ovirt 3.3.2-1.el6 I do not have an option to add a virtio-blk device: Screenshot here: https://dl.dropboxusercontent.com/u/21916057/Screenshot%20from%202014-01-29%... VM disk drive is "Red Hat VirtIO SCSI Disk Device", storage controller is listed as "Red Hat VirtIO SCSI Controller" as shown in device manager. Screenshot here: https://dl.dropboxusercontent.com/u/21916057/Screenshot%20from%202014-01-29%... In Ovirt manager the disk interface is listed as "VirtIO". Screenshot here: https://dl.dropboxusercontent.com/u/21916057/Screenshot%20from%202014-01-29%...

...

...

and my performance is much better. Updates still chug along quite slowly, but I seem to have more than the < 100KB/s write speeds I was seeing previously.

Does anyone know what this device is for? I have the "Red Hat VirtIO SCSI Controller" listed under storage controllers.

[VR] It's a virtio-blk device. OS cannot see this volume unless you have viostor.sys driver installed on it.

Interesting that my VM's can see the controller, but I can't add a disk for that controller in Ovirt. Is there a package I have missed on install? rpm -qa | grep ovirt ovirt-host-deploy-java-1.1.3-1.el6.noarch ovirt-engine-backend-3.3.2-1.el6.noarch ovirt-engine-lib-3.3.2-1.el6.noarch ovirt-engine-restapi-3.3.2-1.el6.noarch ovirt-engine-sdk-python-3.3.0.8-1.el6.noarch ovirt-log-collector-3.3.2-2.el6.noarch ovirt-engine-dbscripts-3.3.2-1.el6.noarch ovirt-engine-webadmin-portal-3.3.2-1.el6.noarch ovirt-host-deploy-1.1.3-1.el6.noarch ovirt-image-uploader-3.3.2-1.el6.noarch ovirt-engine-websocket-proxy-3.3.2-1.el6.noarch ovirt-engine-userportal-3.3.2-1.el6.noarch ovirt-engine-setup-3.3.2-1.el6.noarch ovirt-iso-uploader-3.3.2-1.el6.noarch ovirt-engine-cli-3.3.0.6-1.el6.noarch ovirt-engine-3.3.2-1.el6.noarch ovirt-engine-tools-3.3.2-1.el6.noarch

...

...

I've setup a NFS storage domain on my desktops SSD. I've re-installed win 2008 r2 and initially it was running smoother.

Disk performance peaks at 100MB/s.

If I copy a 250MB file from a share into the Windows VM, it writes out

[VR] Do you copy it with Explorer or any other copy program?

Windows Explorer only.

Do you have HPET enabled?

I can't find it in the guest 'system devices'. On the hosts the current clock source is 'tsc', although 'hpet' is an available option.

How does it work with if you copy from/to local (non-NFS) storage?

Not sure, this is a royal pain to setup. Can I use my ISO domain in two different data centers at the same time? I don't have an option to create an ISO / NFS domain in the local storage DC. When I use the import option with the default DC's ISO domain, I get an error "There is no storage domain under the specified path. Check event log for more details." VDMS logs show "Resource namespace 0e90e574-b003-4a62-867d-cf274b17e6b1_imageNS already registered" so I'm guessing the answer is no. I tried to deploy with WDS, but the 64bit drivers apparently aren't signed, and on x86 I get an error about the NIC not being supported even with the drivers added to WDS.

What is your virtio-win drivers package origin and version?

virtio-win-0.1-74.iso -> http://alt.fedoraproject.org/pub/alt/virtio-win/latest/images/

Thanks, Vadim.

Appreciate it, Steve

On 01/29/2014 07:35 PM, Steve Dainard wrote: ...

In Ovirt 3.3.2-1.el6 I do not have an option to add a virtio-blk device: Screenshot here: https://dl.dropboxusercontent.com/u/21916057/Screenshot%20from%202014-01-29%...

virtio is virtio-blk (in the beginning, there was only one virtio, virtio-blk)

... Do you have HPET enabled?

I can't find it in the guest 'system devices'. On the hosts the current clock source is 'tsc', although 'hpet' is an available option.

How does it work with if you copy from/to local (non-NFS) storage?

Not sure, this is a royal pain to setup. Can I use my ISO domain in two different data centers at the same time? I don't have an option to create an ISO / NFS domain in the local storage DC.

an iso domain can be associated to two data centers (or more, or different engines, etc.)

When I use the import option with the default DC's ISO domain, I get an error "There is no storage domain under the specified path. Check event log for more details." VDMS logs show "Resource namespace 0e90e574-b003-4a62-867d-cf274b17e6b1_imageNS already registered" so I'm guessing the answer is no.

the answer is yes, please open a separate thread on this issue to make it easier to troubleshoot it. thanks

I tried to deploy with WDS, but the 64bit drivers apparently aren't signed, and on x86 I get an error about the NIC not being supported even with the drivers added to WDS.

What is your virtio-win drivers package origin and version?

virtio-win-0.1-74.iso -> http://alt.fedoraproject.org/pub/alt/virtio-win/latest/images/

Thanks, Vadim.

Appreciate it, Steve

I've reconfigured my setup (good succes below, but need clarity on gluster option): Two nodes total, both running virt and glusterfs storage (2 node replica, quorum). I've created an NFS storage domain, pointed at the first nodes IP address. I've launched a 2008 R2 SP1 install with a virtio-scsi disk, and the SCSI pass-through driver on the same node as the NFS domain is pointing at. Windows guest install has been running for roughly 1.5 hours, still "Expanding Windows files (55%) ..." top is showing: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 3609 root 20 0 1380m 33m 2604 S 35.4 0.1 231:39.75 glusterfsd 21444 qemu 20 0 6362m 4.1g 6592 S 10.3 8.7 10:11.53 qemu-kvm This is a 2 socket, 6 core xeon machine with 48GB of RAM, and 6x 7200rpm enterprise sata disks in RAID5 so I don't think we're hitting hardware limitations. dd on xfs (no gluster) time dd if=/dev/zero of=test bs=1M count=2048 2048+0 records in 2048+0 records out 2147483648 bytes (2.1 GB) copied, 4.15787 s, 516 MB/s real 0m4.351s user 0m0.000s sys 0m1.661s time dd if=/dev/zero of=test bs=1k count=2000000 2000000+0 records in 2000000+0 records out 2048000000 bytes (2.0 GB) copied, 4.06949 s, 503 MB/s real 0m4.260s user 0m0.176s sys 0m3.991s I've enabled nfs.trusted-sync ( http://gluster.org/community/documentation/index.php/Gluster_3.2:_Setting_Vo... ) on the gluster volume, and the speed difference is immeasurable . Can anyone explain what this option does, and what the risks are with a 2 node gluster replica volume with quorum enabled? Thanks,

On Fri, 2014-01-31 at 11:37 -0500, Steve Dainard wrote:

I've reconfigured my setup (good succes below, but need clarity on gluster option):

Two nodes total, both running virt and glusterfs storage (2 node replica, quorum).

I've created an NFS storage domain, pointed at the first nodes IP address. I've launched a 2008 R2 SP1 install with a virtio-scsi disk, and the SCSI pass-through driver on the same node as the NFS domain is pointing at.

Windows guest install has been running for roughly 1.5 hours, still "Expanding Windows files (55%) ..."

[VR] Does it work faster with IDE? Do you have kvm enabled? Thanks, Vadim.

top is showing: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND

3609 root 20 0 1380m 33m 2604 S 35.4 0.1 231:39.75 glusterfsd 21444 qemu 20 0 6362m 4.1g 6592 S 10.3 8.7 10:11.53 qemu-kvm

This is a 2 socket, 6 core xeon machine with 48GB of RAM, and 6x 7200rpm enterprise sata disks in RAID5 so I don't think we're hitting hardware limitations.

dd on xfs (no gluster)

time dd if=/dev/zero of=test bs=1M count=2048 2048+0 records in 2048+0 records out 2147483648 bytes (2.1 GB) copied, 4.15787 s, 516 MB/s

real 0m4.351s user 0m0.000s sys 0m1.661s

time dd if=/dev/zero of=test bs=1k count=2000000 2000000+0 records in 2000000+0 records out 2048000000 bytes (2.0 GB) copied, 4.06949 s, 503 MB/s

real 0m4.260s user 0m0.176s sys 0m3.991s

I've enabled nfs.trusted-sync (http://gluster.org/community/documentation/index.php/Gluster_3.2:_Setting_Vo...) on the gluster volume, and the speed difference is immeasurable . Can anyone explain what this option does, and what the risks are with a 2 node gluster replica volume with quorum enabled?

Thanks,