Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

[Tamas Papp]

On 09/02/2013 08:52 AM, Rejy M Cyriac wrote:

On 09/02/2013 12:18 PM, Tamas Papp wrote:

On 09/01/2013 08:10 AM, Bobby Jacob wrote:
  


*From:*Bobby Jacob
*Sent:* Wednesday, August 28, 2013 8:35 AM
*To:* gluster-users@gluster.org 
*Subject:* Backup / Restore for Gluster volumes.

  


Hi,

  


What would be the various options to backup gluster volumes. The
bricks are created with xfs filesystems. I’ve gone through the
concepts of xfsdump concept. Can we schedule daily incremental backups
of the gluster volume or bricks. ??

  


What about restoring the bricks or volumes. ?


I wouldn't make backups under glusterfs. In theory it should work, but
there are too many critical points in the system.

For example you need to stop every transaction (in case of VMs, then all
of them) to make consistent snapshots.
If you have multiple bricks, all of them has to be snapshotted at the
same timeetc.
In addition restoring would be nightmare.

IMHO, it's a crazy idea.


tamas

ui.: I use dirvish (rsync based backup solution) and it works fine,
though cluster side snapshotting feature would help a lot.



http://www.gluster.org/community/documentation/index.php/Features/snapshot


Yes, I know about that, unfortunately it's not ready yet:

3 Current status

 No support availabe for snap'ing a glusterfs volume



tamas
___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users

Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

[Bobby Jacob]

So is it assumed within the community that we can setup a gluster volume, but 
we cannot backup/restore anything. ? Nothing of this sort is being done now. ?

Thanks & Regards,
Bobby Jacob
P SAVE TREES. Please don't print this e-mail unless you really need to.

From: gluster-users-boun...@gluster.org 
[mailto:gluster-users-boun...@gluster.org] On Behalf Of Tamas Papp
Sent: Monday, September 02, 2013 10:10 AM
To: r...@redhat.com
Cc: gluster-users@gluster.org
Subject: Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

On 09/02/2013 08:52 AM, Rejy M Cyriac wrote:

On 09/02/2013 12:18 PM, Tamas Papp wrote:

On 09/01/2013 08:10 AM, Bobby Jacob wrote:





*From:*Bobby Jacob

*Sent:* Wednesday, August 28, 2013 8:35 AM

*To:* gluster-users@gluster.org 


*Subject:* Backup / Restore for Gluster volumes.







Hi,







What would be the various options to backup gluster volumes. The

bricks are created with xfs filesystems. I've gone through the

concepts of xfsdump concept. Can we schedule daily incremental backups

of the gluster volume or bricks. ??







What about restoring the bricks or volumes. ?





I wouldn't make backups under glusterfs. In theory it should work, but

there are too many critical points in the system.



For example you need to stop every transaction (in case of VMs, then all

of them) to make consistent snapshots.

If you have multiple bricks, all of them has to be snapshotted at the

same timeetc.

In addition restoring would be nightmare.



IMHO, it's a crazy idea.





tamas



ui.: I use dirvish (rsync based backup solution) and it works fine,

though cluster side snapshotting feature would help a lot.







http://www.gluster.org/community/documentation/index.php/Features/snapshot

Yes, I know about that, unfortunately it's not ready yet:



3 Current status

 No support availabe for snap'ing a glusterfs volume


tamas
___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users

Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

[Tamas Papp]

Why would you not be able to backup anything? Could you elaborate, what you 
mean?
What do you want to do actually? Maybe we are talking about different things...

tamas

On 09/02/2013 09:54 AM, Bobby Jacob wrote:


So is it assumed within the community that we can setup a gluster volume, but we cannot 
backup/restore anything. ? Nothing of this sort is being done now. ?


Thanks & Regards,

*Bobby Jacob*

P*SAVE TREES**. *Please don't print this e-mail unless you really need to.**

*From:*gluster-users-boun...@gluster.org [mailto:gluster-users-boun...@gluster.org] *On Behalf Of 
*Tamas Papp

*Sent:* Monday, September 02, 2013 10:10 AM
*To:* r...@redhat.com
*Cc:* gluster-users@gluster.org
*Subject:* Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

On 09/02/2013 08:52 AM, Rejy M Cyriac wrote:

On 09/02/2013 12:18 PM, Tamas Papp wrote:

On 09/01/2013 08:10 AM, Bobby Jacob wrote:

  

  


*From:*Bobby Jacob

*Sent:* Wednesday, August 28, 2013 8:35 AM

*To:*gluster-users@gluster.org    
  

*Subject:* Backup / Restore for Gluster volumes.

  

  

  


Hi,

  

  

  


What would be the various options to backup gluster volumes. The

bricks are created with xfs filesystems. I've gone through the

concepts of xfsdump concept. Can we schedule daily incremental 
backups

of the gluster volume or bricks. ??

  

  

  


What about restoring the bricks or volumes. ?

  

  


I wouldn't make backups under glusterfs. In theory it should work, but

there are too many critical points in the system.

  


For example you need to stop every transaction (in case of VMs, then all

of them) to make consistent snapshots.

If you have multiple bricks, all of them has to be snapshotted at the

same timeetc.

In addition restoring would be nightmare.

  


IMHO, it's a crazy idea.

  

  


tamas

  


ui.: I use dirvish (rsync based backup solution) and it works fine,

though cluster side snapshotting feature would help a lot.

  

  

  


http://www.gluster.org/community/documentation/index.php/Features/snapshot


Yes, I know about that, unfortunately it's not ready yet:


3 Current status

  No support availabe for snap'ing a glusterfs volume



tamas



___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users

Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

[Bobby Jacob]

I have 2 servers running gluster. Each server has 1 brick each which is  part 
of a replica-volume. Data is getting replicated onto both the bricks.

* How to backup data from the bricks. ??

* How to backup all gluster configurations. ??

Why I ask this is, we are using this deployment for saving user files. The 
application allots 1GB quota for each user as his personal workspace. Ofcourse, 
we would want to backup the data. !!!

Thanks & Regards,
Bobby Jacob
P SAVE TREES. Please don't print this e-mail unless you really need to.

From: Tamas Papp [mailto:tom...@martos.bme.hu]
Sent: Monday, September 02, 2013 11:44 AM
To: Bobby Jacob
Cc: r...@redhat.com; gluster-users@gluster.org
Subject: Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

Why would you not be able to backup anything? Could you elaborate, what you 
mean?
What do you want to do actually? Maybe we are talking about different things...

tamas

On 09/02/2013 09:54 AM, Bobby Jacob wrote:
So is it assumed within the community that we can setup a gluster volume, but 
we cannot backup/restore anything. ? Nothing of this sort is being done now. ?

Thanks & Regards,
Bobby Jacob
P SAVE TREES. Please don't print this e-mail unless you really need to.

From: 
gluster-users-boun...@gluster.org 
[mailto:gluster-users-boun...@gluster.org] On Behalf Of Tamas Papp
Sent: Monday, September 02, 2013 10:10 AM
To: r...@redhat.com
Cc: gluster-users@gluster.org
Subject: Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

On 09/02/2013 08:52 AM, Rejy M Cyriac wrote:

On 09/02/2013 12:18 PM, Tamas Papp wrote:

On 09/01/2013 08:10 AM, Bobby Jacob wrote:





*From:*Bobby Jacob

*Sent:* Wednesday, August 28, 2013 8:35 AM

*To:* gluster-users@gluster.org 


*Subject:* Backup / Restore for Gluster volumes.







Hi,







What would be the various options to backup gluster volumes. The

bricks are created with xfs filesystems. I've gone through the

concepts of xfsdump concept. Can we schedule daily incremental backups

of the gluster volume or bricks. ??







What about restoring the bricks or volumes. ?





I wouldn't make backups under glusterfs. In theory it should work, but

there are too many critical points in the system.



For example you need to stop every transaction (in case of VMs, then all

of them) to make consistent snapshots.

If you have multiple bricks, all of them has to be snapshotted at the

same timeetc.

In addition restoring would be nightmare.



IMHO, it's a crazy idea.





tamas



ui.: I use dirvish (rsync based backup solution) and it works fine,

though cluster side snapshotting feature would help a lot.







http://www.gluster.org/community/documentation/index.php/Features/snapshot

Yes, I know about that, unfortunately it's not ready yet:




3 Current status

 No support availabe for snap'ing a glusterfs volume


tamas

___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users

Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

[Tamas Papp]

On 09/02/2013 10:51 AM, Bobby Jacob wrote:


I have 2 servers running gluster. Each server has 1 brick each which is  part of a replica-volume. 
Data is getting replicated onto both the bricks.


·How to backup data from the bricks. ??

·How to backup all gluster configurations. ??

Why I ask this is, we are using this deployment for saving user files. The application allots 1GB 
quota for each user as his personal workspace. Ofcourse, we would want to backup the data. !!!





Mount the volume on the backup server and backup files as the would exist 
locally.
Cluster (gluster) configurations can be backed up by either exporting gluster volume info or backing 
up the vol files (/var/lib/glusterfs/... or something like that).



tamas
___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users

[Gluster-users] Rebalance question

[Fred van Zwieten]

Hi,

I have a question. I am on RHS 2.0 Update 4 (will soon go to 2.1 when it's
out) and I have a distributed volume (across 7 nodes) where a fair amount
of directory moves take place (or directory renames). AFAIK this will most
likely give new hash values and so all data needs to be moved around the
pool. Gluster does not do this immediately, but instead creates a link from
the new location to the old location.

Questions:
1. Is there a gluster command to investigate the amount of these links (How
"dirty" is my volume)
2. What command to I use to fix this. I think a "simple" rebalance would do
it.

I would like to create a cron job that checks the amount of links and, if
above a certain level, it will do a rebalance. Alternatively I could do a
fixed periodic rebalance. And I have as yet no idea on what level (what
amount of found links) a rebalance is necessary.

Any thoughts, hints, etc?

Fred

Seeing, contrary to popular wisdom, isn’t believing. It’s where belief
stops, because it isn’t needed any more.. (Terry Pratchett)
___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users

Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

[James]

On Mon, Sep 2, 2013 at 5:17 AM, Tamas Papp  wrote:
> Mount the volume on the backup server and backup files as the would exist
> locally.
I do this using rsnapshot. You might want to have a look, but
something like bacula is another common solution. Because gluster
volumes can be quite large, you might want to have multiple
independent rsnapshot runs working concurrently. I wrote a puppet
module to manage and setup the separate rsnapshot instances which is
not a native rsnapshot feature. Have a look if you like:
https://github.com/purpleidea/puppet-rsnapshot

> Cluster (gluster) configurations can be backed up by either exporting
> gluster volume info or backing up the vol files (/var/lib/glusterfs/... or
> something like that).
Exactly... Or by using something like puppet-gluster to define your
configuration, and then backing up your puppet code. [plug]
https://github.com/purpleidea/puppet-gluster

Hope this helps,
James
___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users

Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

[Fred van Zwieten]

This is an interesting topic. I too am looking for a backup tool that is
"glusterfs friendly". I am looking at rsnapshot, but I want to use
glusterfs as a backup target and rsnapshot relies heavily on hardlinks.
Hardlinks are a special case in glusterfs and I am not sure if rsnapshot is
the right solution for me. I am also looking at Bacula. My use case btw is
tens of thousands of tiff images scatered among directories. These tiff
images sometimes get moved around and even hole subdirectories get moved
around. rsnapshot nor bacula handle that elegantly. snapshotting would, but
that is not build into glusterfs and lvm snapshots are not supported (in
RHS).

Groeten,

Fred

Science flies us to the moon. Religion flies us into buildings (Victor
Stenger)


On Mon, Sep 2, 2013 at 11:33 AM, James  wrote:

> On Mon, Sep 2, 2013 at 5:17 AM, Tamas Papp  wrote:
> > Mount the volume on the backup server and backup files as the would exist
> > locally.
> I do this using rsnapshot. You might want to have a look, but
> something like bacula is another common solution. Because gluster
> volumes can be quite large, you might want to have multiple
> independent rsnapshot runs working concurrently. I wrote a puppet
> module to manage and setup the separate rsnapshot instances which is
> not a native rsnapshot feature. Have a look if you like:
> https://github.com/purpleidea/puppet-rsnapshot
>
> > Cluster (gluster) configurations can be backed up by either exporting
> > gluster volume info or backing up the vol files (/var/lib/glusterfs/...
> or
> > something like that).
> Exactly... Or by using something like puppet-gluster to define your
> configuration, and then backing up your puppet code. [plug]
> https://github.com/purpleidea/puppet-gluster
>
> Hope this helps,
> James
> ___
> Gluster-users mailing list
> Gluster-users@gluster.org
> http://supercolony.gluster.org/mailman/listinfo/gluster-users
>
___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users

Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

[James]

On Mon, Sep 2, 2013 at 6:54 AM, Fred van Zwieten  wrote:
> This is an interesting topic. I too am looking for a backup tool that is
> "glusterfs friendly". I am looking at rsnapshot, but I want to use glusterfs
> as a backup target and rsnapshot relies heavily on hardlinks. Hardlinks are
> a special case in glusterfs and I am not sure if rsnapshot is the right
> solution for me. I am also looking at Bacula. My use case btw is tens of
> thousands of tiff images scatered among directories. These tiff images
> sometimes get moved around and even hole subdirectories get moved around.
> rsnapshot nor bacula handle that elegantly. snapshotting would, but that is
> not build into glusterfs and lvm snapshots are not supported (in RHS).
>
> Groeten,
>
> Fred


My understanding was that hardlinks are slower in gluster because of
additional lookups that are needed to resolve the data. Having said
that, it might still be faster than copying around files, and maybe
someone is working to improve hardlink performance. One of the core
gluster devs will know more about this.

One backup idea could involve having a large gluster volume which
rsnapshots to itself for backup. To restore, you could just hardlink
the files back. I haven't thought about this a lot, but it might be
something to look into. This obviously doesn't help you for offsite
backups in case of a total disaster, so you'll still need to
geo-replicate for safety.

Hopefully this gives you some ideas.

James
___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users

Re: [Gluster-users] Rebalance question

[Fred van Zwieten]

Hi,

>
> I have a question. I am on RHS 2.0 Update 4 (will soon go to 2.1 when it's
> out) and I have a distributed volume (across 7 nodes) where a fair amount
> of directory moves take place (or directory renames). AFAIK this will most
> likely give new hash values and so all data needs to be moved around the
> pool. Gluster does not do this immediately, but instead creates a link from
> the new location to the old location.
>
> Questions:
> 1. Is there a gluster command to investigate the amount of these links
> (How "dirty" is my volume)
> 2. What command to I use to fix this. I think a "simple" rebalance would
> do it.
>
> I would like to create a cron job that checks the amount of links and, if
> above a certain level, it will do a rebalance. Alternatively I could do a
> fixed periodic rebalance. And I have as yet no idea on what level (what
> amount of found links) a rebalance is necessary.
>
> Any thoughts, hints, etc?
>
> Fred
>
> Seeing, contrary to popular wisdom, isn’t believing. It’s where belief
> stops, because it isn’t needed any more.. (Terry Pratchett)
>
___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users

Re: [Gluster-users] FW: Backup / Restore for Gluster volumes.

[Tamas Papp]

On 09/02/2013 12:54 PM, Fred van Zwieten wrote:
This is an interesting topic. I too am looking for a backup tool that is "glusterfs friendly". I 
am looking at rsnapshot, but I want to use glusterfs as a backup target and rsnapshot relies 
heavily on hardlinks. Hardlinks are a special case in glusterfs and I am not sure if rsnapshot is 
the right solution for me. I am also looking at Bacula. My use case btw is tens of thousands of 
tiff images scatered among directories. These tiff images sometimes get moved around and even hole 
subdirectories get moved around. rsnapshot nor bacula handle that elegantly. snapshotting would, 
but that is not build into glusterfs and lvm snapshots are not supported (in RHS).


Why do you depend on glusterfs?

Anyway, rnapshot is good but only in case data is compressible (IMHO).

There are number of similar and free backup solutions and most of them use 
hardlinks.

If you want to keep backup data on a cluster anyhoy, I suggest you to try Lustre, as it's quite good 
in handling metadata.
If you can consider using single (or separatated) server(s) and metadata is still killing your 
server or you need to keep the same or slightly the same files (eg. after chmod actions, 
moving/renaming directories) in backup for multiple times then zfs or btrfs with snapshot feature 
(and rsync --inplace) could be a good choice. In addition compression and/or deduplication can be 
enabled.



tamas
___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users

[Gluster-users] Slow Read Performance With Samba GlusterFS VFS

[kane]

Hi,

First I appreciate for you all, with your last week's points mentioned in the 
problem I meet with slow samba vfs write performance.
Then I use fastcopy software to test windows clients write performance, it 
tells the same performance with iozone rewrite results, reach 450MB/s each 
client.

Then I use the same smb.conf test read performance, it shows slow read 
performance compare with write performance,
I received two clients total result:
500MB/s read perf
900MB/s write perf

I tested the iozone in the mount dir that is mounted with a raid5 block device:
writers = 674063.36 KB/sec 
readers = 502945.33 KB/sec 

then i make the same test in fuse mount dir:
writers = 799379.41 KB/sec 
readers = 898367.59 KB/sec 

it show almost the same performance in read and write, so i was confused with 
the samba vfs client test:
500MB/s read perf
900MB/s write perf

then i want to test vfs use smb.conf sendfile option, but gluster-vfs currently 
not supported.

gluster server volume also set:
performance.read-ahead-page-count: 16
performance.cache-size: 256MB

smb.conf
[global]
workgroup = MYGROUP
server string = DCS Samba Server
log file = /var/log/samba/log.vfs
max log size = 50
#   use sendfile = true
aio read size = 262144
aio write size = 262144
aio write behind = true
min receivefile size = 262144
write cache size = 268435456
security = user
passdb backend = tdbsam
load printers = yes
cups options = raw
read raw = yes
write raw = yes
max xmit = 262144
socket options = TCP_NODELAY IPTOS_LOWDELAY SO_RCVBUF=262144 
SO_SNDBUF=262144
kernel oplocks = no
stat cache = no

So any suggestions tuning the read performance,
Is gluster-vfs aio & sendfile coming soon? 


Thank You!
-kane ___
Gluster-users mailing list
Gluster-users@gluster.org
http://supercolony.gluster.org/mailman/listinfo/gluster-users