[Gluster-users] Debian Jessie (8) qemu-kvm VM and glusterfs as backend storage

Roman romeo.r at gmail.com
Fri May 15 17:13:44 UTC 2015


Yes, I think it is networking related also.
I've tested with ping -f -s 1000 - no losses, but it seems to me, that D8
loses it's connectivity due to some reason. Maybe driver, May be something
else, not the line between servers as I just got the same problem on the
network of 2 switches (the problem I described above appears at the range
of 14 km, but it is only 0,1 ms further, than LAN :) 0,1 ms vs 0,2 ms.
Seems like D8 network driver has changed. Well, I think there is no way to
debug it on acceptable level for Debian guys, so I will wait for 8.1 or
even 8.2. Its too raw anyways. Thanks for your idea.

2015-05-15 19:35 GMT+03:00 Joe Julian <joe at julianfamily.org>:

>  It sounds like a network problem to me. Verify with iperf between your
> client and each server.
>
>
> On 05/15/2015 09:01 AM, Roman wrote:
>
> Actually its pretty random. Now got installed without any problems using
> 3.5.3 .
> don't know what to do. and its only with Debian 8.
>
> 2015-05-15 18:53 GMT+03:00 Roman <romeo.r at gmail.com>:
>
>> Hey,
>>
>>
>>  Seems like the problem exists.
>>
>>  I've got 4 instances of proxmox. I've started to update glusterfs some
>> day to the latest (3.5.3) but I've done the server side and forgot about
>> client side. One of proxmox instances was replaced today and installed the
>> 3.5.3 client. So the problem is between 3.5.3 server and 3.5.3 client. then
>> I've updated glusterfs on other proxmox node and had same problem with d8
>> install.
>>
>>  There is just no go with debian 8 install and qcow file format. If I
>> use raw file format for kvm, installation takes ages.
>>
>>  Need fix asap, as this is you stable gluster version... we are running
>> this on in production.
>>
>> 2015-05-15 16:18 GMT+03:00 Roman <romeo.r at gmail.com>:
>>
>>> Hi,
>>>
>>>  Are there any issues with debian8 and glusterfs?
>>>
>>>
>>>  Got this setup:
>>> 1. promox 3.4
>>> 2. gluster 3.5.3 both server and client (I use gluster repo on proxmox
>>> server)
>>> 3. HA gluster volume
>>>
>>>  What I'm trying to do:
>>> task: create a VM with debian 8, using glusterfs HA volume over 1G
>>> network as a place for VM disk file
>>> result:
>>> a) using raw file format - installation process takes ages (6 hours),
>>> but when finished, VM looks weird: packages installations go slowly
>>> (downloads fast, but installs slow). Same time getting files over internet
>>> gives me usual speed (around 96MB/s, yes, I'm on 1G connection)
>>> b) Using qcow format installation process randomly stops at random step:
>>> getting packages from the repository or installing packages after
>>> donwnloaded them ...
>>>
>>>  c) If i convert VM  with RAW file format to template and then create a
>>> VM using this template, but selecting qcow format, VM runs smoothly.
>>>
>>>  Meanwhile, there are lots of VMs on the same gluster storage (debian7,
>>> centos6, ubuntu) running without any problems.
>>>
>>>  I've started topic on proxmox forum, but devs saying, it should be
>>> upto glusterfs :(
>>>
>>>  Any advice or experiences?
>>>
>>>
>>>  PS
>>> installing deb8 on local storage goes well, without any problems.
>>> installing deb7 on the same setup (same VM even, just another virtual
>>> driver) is ok.
>>> --
>>> Best regards,
>>> Roman.
>>>
>>
>>
>>
>>   --
>> Best regards,
>> Roman.
>>
>
>
>
>  --
> Best regards,
> Roman.
>
>
> _______________________________________________
> Gluster-users mailing listGluster-users at gluster.orghttp://www.gluster.org/mailman/listinfo/gluster-users
>
>
>
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://www.gluster.org/mailman/listinfo/gluster-users
>



-- 
Best regards,
Roman.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150515/02a342ee/attachment.html>


More information about the Gluster-users mailing list