Re: [Gluster-users] 3 Node Replicated Gluster set up as a ring - a good idea?
Will i then get loop? Do i need SPF or something alike? Sorry, i meant STP (Spanning Tree Protocol). Cheers, Mario ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] O_DIRECT (I think) again
Hi, We're using liboindirect and it's works. Regarding not mantained - it's not required bz this is very simple and short lib. 2015-02-13 18:55 GMT+03:00 Anand Avati av...@gluster.org: O_DIRECT support in fuse has been for quite some time now, surely well before 3.4 On Fri, Feb 13, 2015, 02:37 Pedro Serotto pedro.sero...@yahoo.es wrote: Dear All, I am actually using the following software stack: debian wheezy with kernel 3.2.0-4-amd64, glusterfs 3.6.2, openstack Juno, libvirt 1.2.9. If I try to attach a block storage to a running vm, Openstack shows the following error: DeviceIsBusy: The supplied device (vdc) is busy. If I try to attach a block storage to a running vm, Libvirt shows the following error: qemuMonitorTextAddDrive:2621 : operation failed: open disk image file failed Looking up for this issue on the web, I found out that Libvirt tries to open the block device by using O_DIRECT flag on; This last one is supported only by fuse for kernel 3.4. Therefore, I tried to apply some options ( http://www.gluster.org/documentation/use_cases/Virt-store-usecase/) to Gluster, but the problem has not been solved. I also found https://github.com/avati/liboindirect but it is old and not mantained. Does somebody found himself in the same situation? If yes, could you please show me how to solve it by mainteining the same version of my software stack. ThanksRegards Pedro ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users -- Best regards, Artem Kuzmitskiy Severnoe DB JSC ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluster replicate-brick issues (Distrubuted-Replica)
That makes sense, but now that the brick is missing and the data is not replicated, how don I get the brick back in to the trusted pool? Original message From: Subrata Ghosh Date:02/14/2015 9:55 PM (GMT-08:00) To: Thomas Holkenbrink , 'gluster-users@gluster.org' Subject: Re: [Gluster-users] Gluster replicate-brick issues (Distrubuted-Replica) Please try to use commit force option and see whether it works as gluster suggested the other only replace-brick command has been depreciated. [root@Server1 ~]# gluster volume replace-brick $volname $from $to start All replace-brick commands except commit force are deprecated. Do you want to continue? (y/n) y volume replace-brick: success: replace-brick started successfully ID: 0062d555-e7eb-4ebe-a264-7e0baf6e7546 You could try following options , used to resolve our replace-brcik scenario looks like close to your case gluster heal info has some issues. That I requested for a clarification fewdays back. # gluster volume replace-bricks $vol_name $old_brick $new_brick commit force # gluster volume heal $vol_name full #gluster volume $vol_name info -- this shows that the Numer of files=1, even though the file is already healed. Regards, Subrata On 02/14/2015 02:28 PM, Thomas Holkenbrink wrote: We have tried to migrate a Brick from one server to another using the following commands. But the data is NOT being replicated… and the BRICK is not showing up anymore. Gluster still appears to be working but the Bricks are not balanced and I need to add the other Brick for Server3 that I don’t want to do until after Server1:Brick2 gets replicated. This is the command to create the Original Volume: [root@Server1 ~]# gluster volume create Storage1 replica 2 transport tcp Server1:/exp/br01/brick1 Server2:/exp/br01/brick1 Server1:/exp/br02/brick2 Server2:/exp/br02/brick2 This is the Current configuration BEFORE the migration.. Server3 has been Peer Probed successfully but that has been it [root@Server1 ~]# gluster --version glusterfs 3.6.2 built on Jan 22 2015 12:58:11 [root@Server1 ~]# gluster volume status Status of volume: Storage1 Gluster process PortOnline Pid -- Brick Server1:/exp/br01/brick1 49152 Y 2167 Brick Server2:/exp/br01/brick1 49152 Y 2192 Brick Server1:/exp/br02/brick2 49153 Y 2172 --- this is the one that goes missing Brick Server2:/exp/br02/brick2 49153 Y 2193 NFS Server on localhost 2049Y 2181 Self-heal Daemon on localhost N/A Y 2186 NFS Server on Server2 2049Y 2205 Self-heal Daemon on Server2 N/A Y 2210 NFS Server on Server3 2049Y 6015 Self-heal Daemon on Server3 N/A Y 6016 Task Status of Volume Storage1 -- There are no active volume tasks [root@Server1 ~]# gluster volume info Volume Name: Storage1 Type: Distributed-Replicate Volume ID: 9616ce42-48bd-4fe3-883f-decd6c4fcd00 Status: Started Number of Bricks: 2 x 2 = 4 Transport-type: tcp Bricks: Brick1: Server1:/exp/br01/brick1 Brick2: Server2:/exp/br01/brick1 Brick3: Server1:/exp/br02/brick2 Brick4: Server2:/exp/br02/brick2 Options Reconfigured: diagnostics.brick-log-level: WARNING diagnostics.client-log-level: WARNING cluster.entry-self-heal: off cluster.data-self-heal: off cluster.metadata-self-heal: off performance.cache-size: 1024MB performance.cache-max-file-size: 2MB performance.cache-refresh-timeout: 1 performance.stat-prefetch: off performance.read-ahead: on performance.quick-read: off performance.write-behind-window-size: 4MB performance.flush-behind: on performance.write-behind: on performance.io-thread-count: 32 performance.io-cache: on network.ping-timeout: 2 nfs.addr-namelookup: off performance.strict-write-ordering: on [root@Server1 ~]# So we start the Migration of the Brick to the new server using the replace Brick command [root@Server1 ~]# volname=Storage1 [root@Server1 ~]# from=Server1:/exp/br02/brick2 [root@Server1 ~]# to=Server3:/exp/br02/brick2 [root@Server1 ~]# gluster volume replace-brick $volname $from $to start All replace-brick commands except commit force are deprecated. Do you want to continue? (y/n) y volume replace-brick: success: replace-brick started successfully ID: 0062d555-e7eb-4ebe-a264-7e0baf6e7546 [root@Server1 ~]# gluster volume replace-brick $volname $from $to status All replace-brick commands except commit force are deprecated. Do you want to continue? (y/n) y volume replace-brick: success: Number of files migrated = 281 Migration complete At this point everything seems to be in order with no outstanding issues. [root@Server1 ~]# gluster volume status Status of volume: Storage1 Gluster process PortOnline Pid
Re: [Gluster-users] O_DIRECT (I think) again
Yeah, hard to maintain something that's complete and has no bugs. On February 15, 2015 8:20:36 AM PST, Artem Kuzmitskiy artem.kuzmits...@gmail.com wrote: Hi, We're using liboindirect and it's works. Regarding not mantained - it's not required bz this is very simple and short lib. 2015-02-13 18:55 GMT+03:00 Anand Avati av...@gluster.org: O_DIRECT support in fuse has been for quite some time now, surely well before 3.4 On Fri, Feb 13, 2015, 02:37 Pedro Serotto pedro.sero...@yahoo.es wrote: Dear All, I am actually using the following software stack: debian wheezy with kernel 3.2.0-4-amd64, glusterfs 3.6.2, openstack Juno, libvirt 1.2.9. If I try to attach a block storage to a running vm, Openstack shows the following error: DeviceIsBusy: The supplied device (vdc) is busy. If I try to attach a block storage to a running vm, Libvirt shows the following error: qemuMonitorTextAddDrive:2621 : operation failed: open disk image file failed Looking up for this issue on the web, I found out that Libvirt tries to open the block device by using O_DIRECT flag on; This last one is supported only by fuse for kernel 3.4. Therefore, I tried to apply some options ( http://www.gluster.org/documentation/use_cases/Virt-store-usecase/) to Gluster, but the problem has not been solved. I also found https://github.com/avati/liboindirect but it is old and not mantained. Does somebody found himself in the same situation? If yes, could you please show me how to solve it by mainteining the same version of my software stack. ThanksRegards Pedro ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users -- Best regards, Artem Kuzmitskiy Severnoe DB JSC ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users -- Sent from my Android device with K-9 Mail. Please excuse my brevity.___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] 3 Node Replicated Gluster set up as a ring - a good idea?
I think switches are much faster at their job than the kernel and avoiding their use may be a mistake. With regard to gluster, latency affects performance so if it's performance you're looking for, lowest latency switches would be wiser. If it's just network design you're asking for, this might not be the most appropriate forum. On February 14, 2015 2:19:43 AM PST, Ml Ml mliebher...@googlemail.com wrote: Hello List, i would like to build a 3 Node Cluster. I was thinking of such a setup: http://oi58.tinypic.com/2rfgghi.jpg My question: Can i set it up my network that way, that it will still work of one node fails? Just thinking about i want bonding and then i might even need a brige to bring the two sides of the nodes together? Will i then get loop? Do i need SPF or something alike? Could i also use TCP Multipath? ( i have no experience with tcp multipath so far) What do you think? Thanks, Mario p.s. i want to avoid switches here. ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users -- Sent from my Android device with K-9 Mail. Please excuse my brevity.___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluster replicate-brick issues (Distrubuted-Replica)
Of those missing files, are they maybe dht link files? Mode 1000, size 0. On February 14, 2015 12:58:12 AM PST, Thomas Holkenbrink thomas.holkenbr...@fibercloud.com wrote: We have tried to migrate a Brick from one server to another using the following commands. But the data is NOT being replicated... and the BRICK is not showing up anymore. Gluster still appears to be working but the Bricks are not balanced and I need to add the other Brick for Server3 that I don't want to do until after Server1:Brick2 gets replicated. This is the command to create the Original Volume: [root@Server1 ~]# gluster volume create Storage1 replica 2 transport tcp Server1:/exp/br01/brick1 Server2:/exp/br01/brick1 Server1:/exp/br02/brick2 Server2:/exp/br02/brick2 This is the Current configuration BEFORE the migration.. Server3 has been Peer Probed successfully but that has been it [root@Server1 ~]# gluster --version glusterfs 3.6.2 built on Jan 22 2015 12:58:11 [root@Server1 ~]# gluster volume status Status of volume: Storage1 Gluster process PortOnline Pid -- Brick Server1:/exp/br01/brick1 49152 Y 2167 Brick Server2:/exp/br01/brick1 49152 Y 2192 Brick Server1:/exp/br02/brick2 49153 Y 2172 --- this is the one that goes missing Brick Server2:/exp/br02/brick2 49153 Y 2193 NFS Server on localhost 2049Y 2181 Self-heal Daemon on localhost N/A Y 2186 NFS Server on Server2 2049Y 2205 Self-heal Daemon on Server2 N/A Y 2210 NFS Server on Server3 2049Y 6015 Self-heal Daemon on Server3 N/A Y 6016 Task Status of Volume Storage1 -- There are no active volume tasks [root@Server1 ~]# gluster volume info Volume Name: Storage1 Type: Distributed-Replicate Volume ID: 9616ce42-48bd-4fe3-883f-decd6c4fcd00 Status: Started Number of Bricks: 2 x 2 = 4 Transport-type: tcp Bricks: Brick1: Server1:/exp/br01/brick1 Brick2: Server2:/exp/br01/brick1 Brick3: Server1:/exp/br02/brick2 Brick4: Server2:/exp/br02/brick2 Options Reconfigured: diagnostics.brick-log-level: WARNING diagnostics.client-log-level: WARNING cluster.entry-self-heal: off cluster.data-self-heal: off cluster.metadata-self-heal: off performance.cache-size: 1024MB performance.cache-max-file-size: 2MB performance.cache-refresh-timeout: 1 performance.stat-prefetch: off performance.read-ahead: on performance.quick-read: off performance.write-behind-window-size: 4MB performance.flush-behind: on performance.write-behind: on performance.io-thread-count: 32 performance.io-cache: on network.ping-timeout: 2 nfs.addr-namelookup: off performance.strict-write-ordering: on [root@Server1 ~]# So we start the Migration of the Brick to the new server using the replace Brick command [root@Server1 ~]# volname=Storage1 [root@Server1 ~]# from=Server1:/exp/br02/brick2 [root@Server1 ~]# to=Server3:/exp/br02/brick2 [root@Server1 ~]# gluster volume replace-brick $volname $from $to start All replace-brick commands except commit force are deprecated. Do you want to continue? (y/n) y volume replace-brick: success: replace-brick started successfully ID: 0062d555-e7eb-4ebe-a264-7e0baf6e7546 [root@Server1 ~]# gluster volume replace-brick $volname $from $to status All replace-brick commands except commit force are deprecated. Do you want to continue? (y/n) y volume replace-brick: success: Number of files migrated = 281 Migration complete At this point everything seems to be in order with no outstanding issues. [root@Server1 ~]# gluster volume status Status of volume: Storage1 Gluster process PortOnline Pid -- Brick Server1:/exp/br01/brick1 49152 Y 2167 Brick Server2:/exp/br01/brick1 49152 Y 2192 Brick Server1:/exp/br02/brick2 49153 Y 27557 Brick Server2:/exp/br02/brick2 49153 Y 2193 NFS Server on localhost 2049Y 27562 Self-heal Daemon on localhost N/A Y 2186 NFS Server on Server2 2049Y 2205 Self-heal Daemon on Server2 N/A Y 2210 NFS Server on Server3 2049Y 6015 Self-heal Daemon on Server3 N/A Y 6016 Task Status of Volume Storage1 -- Task : Replace brick ID : 0062d555-e7eb-4ebe-a264-7e0baf6e7546 Source Brick : Server1:/exp/br02/brick2 Destination Brick: Server3:/exp/br02/brick2 Status : completed The volume reports that the replace Brick command completed.. so the next step is to commit the change [root@Server1 ~]# gluster volume replace-brick $volname $from $to commit All replace-brick commands except commit force are deprecated. Do you
[Gluster-users] basic architecture
Something I don't understand. If I have two servers running gluster code, each with an attached brick, is there a master slave relationship or are they peers? Supposing a have my brick mounted as a document root on each of the two servers, and am running an app, such as Wordpress, or Magento behind a load balancer: * If one goes down, and the other one continues, what happens when I bring the first one back? Does the brick rebuild itself on the downed server? * What if I want to bring a third one into the mix? It comes up with an empty, or obsolete, brick. How does the third brick get in sync? Thanks, Ed Greenberg ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
[Gluster-users] Migrating from files to libgfapi
Hello, I have been looking for a while, but I can not find the proper procedure. I have got some VMs stored in a shared folder distributed using NFS. Additionally, I am switching our storage to GlusterFS, and I plan to use native access using libgfapi in libvirt. Is there a way how to import such a raw files to glusterfs such that it was possible to use them using libgfapi? Thank you, Milos ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] basic architecture
On 02/16/2015 04:20 AM, Ed Greenberg wrote: Something I don't understand. If I have two servers running gluster code, each with an attached brick, is there a master slave relationship or are they peers? There is no master slave relationship here. Each brick can be considered as individual storage entity. Supposing a have my brick mounted as a document root on each of the two servers, and am running an app, such as Wordpress, or Magento behind a load balancer: * If one goes down, and the other one continues, what happens when I bring the first one back? Does the brick rebuild itself on the downed server? If you have set up a replicated volume then once the brick comes back, self heal will get kicked in and it will try to come in sync with its replica pair. * What if I want to bring a third one into the mix? It comes up with an empty, or obsolete, brick. How does the third brick get in sync? I believe you are talking about add-brick command here. In that case at the of provisioning it is ideally empty (we recommend to be it as empty) and then once you kick in rebalance, this brick will also get load balanced along with other bricks in the cluster. Thanks, Ed Greenberg ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users -- ~Atin ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
[Gluster-users] Changelog dir in gluster 3.6
Hi, I don't understand if the changelog is needed to replica and geo-replica situations, so the filesystem is fill of a lot of that files in /.glusterfs/changelogs. Is there anyway to reduce or waive directory? Thanks. ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Warning: W [socket.c:611:__socket_rwv] 0-management: readv on
On 02/16/2015 12:37 PM, Félix de Lelelis wrote: Hi, The last week upgraded us cluster to 3.6 version. I noticed in the log the following error: W [socket.c:611:__socket_rwv] 0-management: readv on /var/run/f3fcde54ca5d30115274155a37baa079.socket failed (Invalid argument) It is due a nfs daemon? It could be, check whether nfs daemon is running through volume status command. If not then look at the nfs.log, most probably the nfs kernel module is not disabled. ~Atin Thanks. ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users -- ~Atin ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluster performance on the small files
Hi, My VM disk are based on RAW... On Sat, Feb 14, 2015 at 4:08 PM, Samuli Heinonen samp...@neutraali.net wrote: Hi! What image type you are using to store virtual machines? For example using sparse QCOW2 images is much slower than preallocated RAW images. Performance with QCOW2 should get better after image file has grown bigger and it's not necessary to resize sparse image anymore. Best regards, Samuli Heinonen On 13.2.2015, at 8.58, Punit Dambiwal hypu...@gmail.com wrote: Hi, I have seen the gluster performance is dead slow on the small files...even i am using the SSDit's too bad performanceeven i am getting better performance in my SAN with normal SATA disk... I am using distributed replicated glusterfs with replica count=2...i have all SSD disks on the brick... root@vm3:~# dd bs=64k count=4k if=/dev/zero of=test oflag=dsync 4096+0 records in 4096+0 records out 268435456 bytes (268 MB) copied, 57.3145 s, 4.7 MB/s root@vm3:~# dd bs=64k count=4k if=/dev/zero of=test conv=fdatasync 4096+0 records in 4096+0 records out 268435456 bytes (268 MB) copied, 1.80093 s, 149 MB/s Thanks, Punit ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
[Gluster-users] Warning: W [socket.c:611:__socket_rwv] 0-management: readv on
Hi, The last week upgraded us cluster to 3.6 version. I noticed in the log the following error: W [socket.c:611:__socket_rwv] 0-management: readv on /var/run/f3fcde54ca5d30115274155a37baa079.socket failed (Invalid argument) It is due a nfs daemon? Thanks. ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] gluster volume heal info giving wrong report
AFR team (Pranith/Ravi cced), FYA.. ~Atin On 02/13/2015 07:48 PM, Subrata Ghosh wrote: Hi All, Can any one clarify the issue e , we are facing with heal incorrect report mentioned below , We are using gluster 3.3.2 . *Issue:* *Bug 1039544* https://bugzilla.redhat.com/show_bug.cgi?id=1039544 -[FEAT] gluster volume heal info should list the entries that actually required to be healed. *Steps to recreate* If we are continuously writing to a file in a volume. At the same time if we do the following # gluster volume replace-bricks $vol_name $old_brick $new_brick commit force # gluster volume heal $vol_name full #gluster volume $vol_name info -- this shows that the Numer of files=1, even though the file is already healed. *Related issues:* http://comments.gmane.org/gmane.comp.file-systems.gluster.user/14188 *Fix:* http://review.gluster.org/#/c/7480/ *Query:* Is the fix that we are trying to back port to gluster 3.2 is the correct code changes. Regards, Subrata ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users -- ~Atin ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] read only feature only works after the volume is restarted?
On 02/13/2015 02:05 PM, Feng Wang wrote: Hi all, If we set the read-only feature using the following command in the cli to a volume in service, it will not work until the volume is restarted. That's the correct functionality. http://review.gluster.org/#/c/8571/ should address it, however it hasn't been taken yet. gluster volume set vol-name features.readonly on It means that the service must be stopped temporarily. Does this make sense? An alternative method is re-mounting the glusterfs using the -o ro option. So both of them need the service to be interrupted . Could we have the method to make the volume read-only after flushing the requests on the fly without stopping the service? Best Regards, Fang Huang ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users -- ~Atin ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users