[Gluster-devel] gfapi 3.6.3 QEMU 2.3 Ubuntu 14.04 testing
Josh Boon
gluster at joshboon.com
Wed Jul 29 14:45:02 UTC 2015
Hey Prasanna,
Thanks for your help! One of the issues we've had is DD doesn't seem to reproduce it. Anything that logs and handles large volumes, think mail and web servers, tends to segfault the most frequently. I could write up a load test and we could put apache on it and try that as that's closet to what we run. Also if you don't object would I be able to get on the machine to figure out apparmor and do a writeup? Most folks probably won't be able to disable it completely.
Best,
Josh
----- Original Message -----
From: "Prasanna Kalever" <pkalever at redhat.com>
To: "Josh Boon" <gluster at joshboon.com>
Cc: "Pranith Kumar Karampuri" <pkarampu at redhat.com>, "Gluster Devel" <gluster-devel at gluster.org>
Sent: Wednesday, July 29, 2015 1:54:34 PM
Subject: Re: [Gluster-devel] gfapi 3.6.3 QEMU 2.3 Ubuntu 14.04 testing
Hi Josh Boon,
Below are my setup details:
# qemu-system-x86_64 --version
QEMU emulator version 2.3.0 (Debian 1:2.3+dfsg-5ubuntu4), Copyright (c) 2003-2008 Fabrice Bellard
# gluster --version
glusterfs 3.6.3 built on Jul 29 2015 16:01:10
Repository revision: git://git.gluster.com/glusterfs.git
Copyright (c) 2006-2011 Gluster Inc. <http://www.gluster.com>
GlusterFS comes with ABSOLUTELY NO WARRANTY.
# lsb_release -a
You may redistribute copies of GlusterFS under the terms of the GNU General Public License.
Distributor ID: Ubuntu
Description: Ubuntu 14.04 LTS
Release: 14.04
Codename: trusty
# gluster vol info
Volume Name: vol1
Type: Replicate
Volume ID: ad78ac6c-c55e-4f4a-8b1b-a11865f1d01e
Status: Started
Number of Bricks: 1 x 2 = 2
Transport-type: tcp
Bricks:
Brick1: 10.70.1.156:/brick1
Brick2: 10.70.1.156:/brick2
Options Reconfigured:
server.allow-insecure: on
storage.owner-uid: 116
storage.owner-gid: 125
#gluster vol status
Status of volume: vol1
Gluster process Port Online Pid
------------------------------------------------------------------------------
Brick 10.70.1.156:/brick1 49152 Y 3726
Brick 10.70.1.156:/brick2 49153 Y 7014
NFS Server on localhost 2049 Y 7028
Self-heal Daemon on localhost N/A Y 7035
Task Status of Volume vol1
------------------------------------------------------------------------------
There are no active volume tasks
# du -sh /brick1/vm1.img
8.6G /brick1/vm1.img
# du -sh /brick2/vm1.img
8.6G /brick2/vm1.img
# cat vm1.xml
...
<devices>
<emulator>/usr/bin/qemu-system-x86_64</emulator>
<disk type='network' device='disk'>
<driver name='qemu' type='qcow2' cache='none'/>
<source protocol='gluster' name='vol1/vm1.img'>
<host name='10.70.1.156' port='24007'/>
</source>
<target dev='vda' bus='virtio'/>
<alias name='virtio-disk0'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x04' function='0x0'/>
</disk>
<disk type='file' device='cdrom'>
<driver name='qemu' type='raw'/>
<source file='/home/pkalever/work/qemu/ubuntu-14.04-server-amd64.iso'/>
<target dev='hdb' bus='ide'/>
<readonly/>
<alias name='ide0-0-1'/>
<address type='drive' controller='0' bus='0' target='0' unit='1'/>
</disk>
...
</devices>
With the underline setup given above, Once the VM booted successfully
I have wrote 5G using dd command but I have not encountered any crashes
I feel qemu- 2.3.0 doesn't have the segfaults issue.
Please revert back for any information if required.
Thanks & regards,
Prasanna Kumar K.
----- Original Message -----
From: "Prasanna Kalever" <pkalever at redhat.com>
To: "Josh Boon" <gluster at joshboon.com>
Cc: "Pranith Kumar Karampuri" <pkarampu at redhat.com>, "SATHEESARAN" <sasundar at redhat.com>
Sent: Tuesday, July 28, 2015 5:04:46 PM
Subject: Re: [Gluster-devel] gfapi 3.6.3 QEMU 2.3 Ubuntu 14.04 testing
Hi Josh Boon,
finally able to boot the VM
for now I disabled apparmor by update-rc.d -f apparmor remove
Thanks for the support, Now I shall try on reproducing the actual problem :)
Best Regards,
Prasanna Kumar K.
----- Original Message -----
From: "Prasanna Kalever" <pkalever at redhat.com>
To: "Josh Boon" <gluster at joshboon.com>
Cc: "Pranith Kumar Karampuri" <pkarampu at redhat.com>, "SATHEESARAN" <sasundar at redhat.com>
Sent: Tuesday, July 28, 2015 3:45:12 PM
Subject: Re: [Gluster-devel] gfapi 3.6.3 QEMU 2.3 Ubuntu 14.04 testing
Hi Josh Boon,
apparmor trick didn't work,
How ever I can notice some apparmor="DENIED" logs (even after adding your lines in /etc/apparmor.d/abstractions/libvirt-qemu)
Jul 27 11:46:16 dhcp-0-156 kernel: [ 1152.096873] type=1400 audit(1437977776.500:55): apparmor="DENIED" operation="open" profile="libvirt-331cf08a-c120-45fb-92fe-c5369c8fdf12" name="/sys/devices/system/node/" pid=3094 comm="qemu-system-x86" requested_mask="r" denied_mask="r" fsuid=0 ouid=0
Jul 27 11:46:16 dhcp-0-156 kernel: [ 1152.096902] type=1400 audit(1437977776.500:56): apparmor="DENIED" operation="open" profile="libvirt-331cf08a-c120-45fb-92fe-c5369c8fdf12" name="/sys/devices/system/cpu/" pid=3094 comm="qemu-system-x86" requested_mask="r" denied_mask="r" fsuid=0 ouid=0
Jul 27 11:46:16 dhcp-0-156 kernel: [ 1152.098569] type=1400 audit(1437977776.500:57): apparmor="DENIED" operation="file_mmap" profile="libvirt-331cf08a-c120-45fb-92fe-c5369c8fdf12" name="/usr/lib/x86_64-linux-gnu/qemu/block-curl.so" pid=3094 comm="qemu-system-x86" requested_mask="m" denied_mask="m" fsuid=0 ouid=107
Jul 27 11:46:16 dhcp-0-156 kernel: [ 1152.098681] type=1400 audit(1437977776.500:58): apparmor="DENIED" operation="file_mmap" profile="libvirt-331cf08a-c120-45fb-92fe-c5369c8fdf12" name="/usr/lib/x86_64-linux-gnu/qemu/block-rbd.so" pid=3094 comm="qemu-system-x86" requested_mask="m" denied_mask="m" fsuid=0 ouid=107
Jul 27 11:46:16 dhcp-0-156 kernel: [ 1152.098733] type=1400 audit(1437977776.500:59): apparmor="DENIED" operation="file_mmap" profile="libvirt-331cf08a-c120-45fb-92fe-c5369c8fdf12" name="/usr/lib/x86_64-linux-gnu/qemu/block-gluster.so" pid=3094 comm="qemu-system-x86" requested_mask="m" denied_mask="m" fsuid=0 ouid=107
I am still working on it, please find the logs attached.
Sorry for the delay, I was working on something else in parallel.
Thanks & Regards,
Prasanna Kumar K.
----- Original Message -----
From: "Josh Boon" <gluster at joshboon.com>
To: "Prasanna Kalever" <pkalever at redhat.com>
Cc: "Pranith Kumar Karampuri" <pkarampu at redhat.com>, "SATHEESARAN" <sasundar at redhat.com>
Sent: Thursday, July 23, 2015 9:42:56 PM
Subject: Re: [Gluster-devel] gfapi 3.6.3 QEMU 2.3 Ubuntu 14.04 testing
Hi Prasanna,
You've hit apparmor. I should work on providing better instructions for folks over at gluster.org :)
To fix up apparmor:
edit /etc/apparmor.d/abstractions/libvirt-qemu
and include at the top of the file
# For gluster use
/usr/lib/x86_64-linux-gnu/glusterfs/** rmix,
/proc/sys/net/ipv4/ip_local_reserved_ports r,
/tmp/** rwcx,
and restart apparmor
service apparmor restart
if you still have issues please send me the output of your syslog and qemu-img --help
----- Original Message -----
From: "Prasanna Kalever" <pkalever at redhat.com>
To: "Josh Boon" <gluster at joshboon.com>
Cc: "Pranith Kumar Karampuri" <pkarampu at redhat.com>, "SATHEESARAN" <sasundar at redhat.com>
Sent: Thursday, July 23, 2015 12:20:33 PM
Subject: Re: [Gluster-devel] gfapi 3.6.3 QEMU 2.3 Ubuntu 14.04 testing
Hi Josh Boon,
Thanks for your reply
To avoid issues caused by nested vm's now I moved to to setup
1. Native Ubuntu-14.04
2. with qemu from ppa provided
3. compiled from source gluster-3.6.3
To set up VM's, I followed your procedure modified VARRAY, volume name and IP, but still encountering some issues,
# virsh start HFM
error: Failed to start domain HFM
error: internal error: process exited while connecting to monitor: Failed to open module: /usr/lib/x86_64-linux-gnu/qemu/block-curl.so: failed to map segment from shared object: Permission denied
Failed to open module: /usr/lib/x86_64-linux-gnu/qemu/block-rbd.so: failed to map segment from shared object: Permission denied Failed to open module: /usr/lib/x86_64-linux-gnu/qemu/block-gluster.so: failed to map segment from shared object: Permission denied
qemu-system-x86_64: -drive file=gluster://10.70.1.156/mnt/vm1_mount//HFM.img,if=none,id=drive-virtio-disk0,format=raw,cache=none: Unknown protocol 'gluster'
Best Regards,
Prasanna Kumar K.
----- Original Message -----
From: Josh Boon <gluster at joshboon.com>
To: Prasanna Kalever <pkalever at redhat.com>
Cc: Pranith Kumar Karampuri <pkarampu at redhat.com>
Sent: Mon, 20 Jul 2015 09:02:32 -0400 (EDT)
Subject: Re: [Gluster-devel] gfapi 3.6.3 QEMU 2.3 Ubuntu 14.04 testing
I'm not familiar withe the virt-install method. Can you provide me with the output of virsh dumpxml VMubuntu_64?
I've attached my template XML I use and the seed files necessary to generate an ISO for ubuntu's cloud images if you'd like to go that route. Steps follow for use
1. modify template.xml for gluster array. I'm using 10.9.1.1 and VMARRAY as the name of my gluster volume. Also don't forget to modify the processor if unsupported and the hostname. define the machine
2. Build out a preseed iso (or not.) The two *data files attached recreate the cloud-init template necessary for a self-contained install. None of it is absolutely necessary but can help make your stay on the server more pleasant. Change values as necessary and then generate the instance id and iso image
2a. apt-get install genisoimage; echo "instance-id: $(uuidgen || echo i-abcdefg)" >> meta-data ; genisoimage -output /mnt/VMARRAY/$IMAGE-seed.iso -volid cidata -joliet -rock meta-data user-data
3. virsh start --console $MACHINE
----- Original Message -----
From: "Prasanna Kalever" <pkalever at redhat.com>
To: "Josh Boon" <gluster at joshboon.com>
Cc: "Pranith Kumar Karampuri" <pkarampu at redhat.com>
Sent: Monday, July 13, 2015 9:48:08 AM
Subject: RE: [Gluster-devel] gfapi 3.6.3 QEMU 2.3 Ubuntu 14.04 testing
Hi Josh Boon,
I am facing issue ("ERROR XML error: No PCI buses available") while # virsh start <VM>
set up details:
---------------
Native Install Fedora21
Running "Ubuntu amd64 trusty 14.04.02 LTS" as a VM using virt-manager in fedora21.
Installed qemu and libseccomp form the ppa http://ppa.launchpad.net/josh-boon/qemu-edge-glusterfs/ubuntu and others form trusty repo
Issue caused while starting a level 2 VM in Ubuntu-14.04 (which is again a VM in fedora)
steps used:
----------
# qemu-img create -f qcow2 gluster://IP/VOLNAME/vm1_file.img 15G
Successful
# mount gluster volume
successful
# virt-install -n VMubuntu_64 -r 256 --disk path=/mnt/vm1_mount/vm1_file.img,bus=virtio,size=4 -c /home/pkalever/ubuntu-14.04-server-amd64.iso --network network=default,model=virtio --nographics --accelerate
WARNING KVM acceleration not available, using 'qemu'
Starting install...
ERROR XML error: No PCI buses available
Domain installation does not appear to have been successful.
If it was, you can restart your domain by running:
virsh --connect qemu:///system start VMubuntu_64
otherwise, please restart your installation.
Other details:
--------------
Following commands executed in Level 1 VM (Ubuntu 14.04 amd64)
# qemu-x86_64 -version
qemu-x86_64 version 2.3.0 (Debian 1:2.3+dfsg-5ubuntu4), Copyright (c) 2003-2008 Fabrice Bellard
# glusterfs --version
glusterfs 3.6.3 built on Jul 9 2015 15:51:55
Repository revision: git://git.gluster.com/glusterfs.git
...
# libvirtd --version
libvirtd (libvirt) 1.2.2
# uname -a
Linux root 3.13.0-24-generic #46-Ubuntu SMP Thu Apr 10 19:11:08 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux
# lsb_release -a
Distributor ID: Ubuntu
Description: Ubuntu 14.04.2 LTS
Release: 14.04
Codename: trusty
I am expecting this issue might be caused due to 2 level nested VM's creation.
Please revert back for in case of any other information regarding the setup.
Best Regards,
Prasanna Kumar K.
----- Forwarded Message -----
From: "Pranith Kumar Karampuri" <pkarampu at redhat.com>
To: "Josh Boon" <gluster at joshboon.com>, gluster-infra at gluster.org, "Gluster Devel" <gluster-devel at gluster.org>, "Prasanna Kalever" <pkalever at redhat.com>
Sent: Thursday, July 9, 2015 11:02:06 AM
Subject: Re: [Gluster-devel] gfapi 3.6.3 QEMU 2.3 Ubuntu 14.04 testing
CC Prasanna who will be looking into it.
On 07/06/2015 07:30 PM, Josh Boon wrote:
Hey folks,
Does anyone have test environment running Ubuntu 14.04, QEMU 2.0, and Gluster 3.6.3? I'm looking to have some folks test out QEMU 2.3 for stability and performance and see if it removes the segfault errors. Another group of folks are experiencing the same segfaults I still experience but looking over their logs my theory of it being related to a self-heal didn't work out. I've included the stack trace below from their environment which matches mine. I've already put together a PPA over at https://launchpad.net/~josh-boon/+archive/ubuntu/qemu-edge-glusterfs with QEMU 2.3 and deps built for trusty. If anyone has the time or the resources that I could get into I'd appreciate the support. I'd like to get this ironed out so I can give my full vote of confidence to Gluster again.
Thanks,
Josh
Stack
#0 0x00007f369c95248c in ?? ()
No symbol table info available.
#1 0x00007f369bd2b3b1 in glfs_io_async_cbk (ret=<optimized out>, frame=<optimized out>, data=0x7f369ee536c0) at glfs-fops.c:598
gio = 0x7f369ee536c0
#2 0x00007f369badb66a in syncopctx_setfspid (pid=0x7f369ee536c0) at syncop.c:191
opctx = 0x0
ret = -1
#3 0x0000000000100011 in ?? ()
No symbol table info available.
#4 0x00007f36a5ae26b0 in ?? ()
No symbol table info available.
#5 0x00007f36a81e2800 in ?? ()
No symbol table info available.
#6 0x00007f36a5ae26b0 in ?? ()
No symbol table info available.
#7 0x00007f36a81e2800 in ?? ()
No symbol table info available.
#8 0x0000000000000000 in ?? ()
No symbol table info available.
Full log attached.
_______________________________________________
Gluster-devel mailing list Gluster-devel at gluster.org http://www.gluster.org/mailman/listinfo/gluster-devel
_______________________________________________
Gluster-devel mailing list
Gluster-devel at gluster.org
http://www.gluster.org/mailman/listinfo/gluster-devel
More information about the Gluster-devel
mailing list