[Gluster-users] Slow self-heal while increasing replica
Hello, I am in the process of increasing the replica from 4x2 to 4x3 bricks on three physical nodes; even if the three hosts have 4x1 Gb ethernet interfaces connected to the same switch with "balance-alb" bonding, I am seeing extreme slowness in replicating the data to the third replica, somewhere near 40-50 Mbit/s. I have tested the bandwidth using netperf and proved that it can reach 800-900 Mbit/s per channel (that is, per source host up to 4 hosts). Is there any GlusterFS option I should check to verify there are no bottlenecks? I am still using Gluster 3.4.7 because of the following: - https://bugzilla.redhat.com/show_bug.cgi?id=1168897 "remove-brick" reports "Cluster op-version must atleast be 30600." - https://bugzilla.redhat.com/show_bug.cgi?id=1113778 gluster volume heal info reports "Volume heal failed" Regards ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
[Gluster-users] Expanding gluster by one server
Hello, I'll soon be trying to implement the following scenario: https://joejulian.name/blog/how-to-expand-glusterfs-replicated-clusters-by-one-server/ Instead of the deprecated "replace-brick" command, can I temporarily increase the replica number to 3, wait for data to propagate, and after that go back to replica 2 with different "brick" sets? I think this could be convenient because I currently have two servers with four bricks *each* and replica 2. I have no spare space left on the servers, so I'm afraid the procedure described at http://www.gluster.org/pipermail/gluster-users/2013-September/014537.html cannot be used in my case. Regards -- https://s19n.net ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] GlusterFS 3.4.0 and 3.3.2 released!
* Vijay Bellur [2013 07 18, 00:00]: > There are no known problems with ext4 and 3.4.0, even with newer kernels. What about 3.3.2? Does the same apply for 3.3.2? Regards ___ Gluster-users mailing list Gluster-users@gluster.org http://supercolony.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] gluster fs hangs on certain operations
* Scott Hazelhurst [2013 03 15, 11:49]: > We are running gluster 3.3.1 on SL 6.3. The bricks are formatted ext3 http://joejulian.name/blog/glusterfs-bit-by-ext4-structure-change/ ? Maybe the solution is "use xfs". Regards ___ Gluster-users mailing list Gluster-users@gluster.org http://supercolony.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Gluster machines slowing down over time
> > On Tue, Dec 11, 2012 at 04:13:15PM +, Tom Hall wrote: > > I notice when sshing into the machine takes ages and running remote commands > > with capistrano takes longer and longer. That is commonly related to DNS unavailability or misconfiguration. What makes you think the slowness is gluster-related? Regards ___ Gluster-users mailing list Gluster-users@gluster.org http://supercolony.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] worth upgrading from 3.2.7 to 3.3.1?
* Gerald Brandt [2012 11 27, 09:36]: > I have speed and stability increases with 3.3.0/3.3.1. Sorry if this may seem thread hijacking, but have you had success in mixing the two versions (3.3.0 and 3.3.1) in the same cluster? Is then possible to perform a rolling upgrade with no downtime? I already posted this very same question on Oct 22nd, but it's like it's so obvious that no-one needs to answer, though I can't find the answer anywhere in the documentation. Best regards ___ Gluster-users mailing list Gluster-users@gluster.org http://supercolony.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] upgrade from 3.3.0 to 3.3.1
* Patrick Irvine [2012 10 16, 22:03]: > Can I do a simple rolling upgrade from 3.3.0 to 3.3.1? Or are > there some gotchas? I haven't seen any answer to this enquiry, though I think it is very simple and important to know the answer. Can 3.3.0/3.3.1 bricks coexist? What about clients, can 3.3.1 clients connect to 3.3.0 bricks? Thank you very much in advance for your answers, Best regards ___ Gluster-users mailing list Gluster-users@gluster.org http://supercolony.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] rdma tansport on 3.3?
* samuel [2012 10 19, 11:44]: > I've tried 3.3.1, 3.4-rc1,3.2.7 and only been able to mount the volume > with 3.2.7. With the other gluster versions, the following error appears > on the gluster logs: In the Admin-Guide the following can be read (taken from 3.3.1 changelog): Command Reference - The Gluster Console Manager "NOTE: with 3.3.0 release, transport type 'rdma' and 'tcp,rdma' are not fully supported" Regards ___ Gluster-users mailing list Gluster-users@gluster.org http://supercolony.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Clearing the heal-failed and split-brain status messages
* Robert van Leeuwen [2012 10 10, 13:56]: > > From: gluster-users-boun...@gluster.org > > If this were a "wishlist" bug on the bugzilla tracker, I'd subscribe to > > it (which seems the common practice to show interest and say "+1"). > > I've created an issue on the bugtracker: > https://bugzilla.redhat.com/show_bug.cgi?id=864963 Thanks, subscribed. Regards ___ Gluster-users mailing list Gluster-users@gluster.org http://supercolony.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Clearing the heal-failed and split-brain status messages
* Robert van Leeuwen [2012 10 10, 08:39]: > Is it possible to clear the heal-failed and split-brain status in a nice way? > I would personally like if gluster would automatically remove failed states > when they are resolved ( if future reference is needed you can always look at > the logs) If this were a "wishlist" bug on the bugzilla tracker, I'd subscribe to it (which seems the common practice to show interest and say "+1"). Regards ___ Gluster-users mailing list Gluster-users@gluster.org http://supercolony.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Tuning/Configuration: Specifying the TCP port opened by glusterfsd for each brick?
* Eric [2012 09 24, 20:08]: > Anybody? > It's been three weeks and nobody's even made a suggestion! Have you tried searching the mailing list archives during these three weeks? http://gluster.org/pipermail/gluster-users/2011-February/006709.html "Currently, there is no option to change the port through options, hence you need to change the code." Regards ___ Gluster-users mailing list Gluster-users@gluster.org http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
Re: [Gluster-users] New release of Gluster?
* Gerald Brandt [2012 09 18, 11:16]: > Are there any proposed dates for a new release of Gluster? > I'm currently running 3.3, and the gluster heal info commands > all segfault. It happened to me as well. I'd support the request for a new release, which could also address the known problems with ext4... For what relates to the segfaults: https://bugzilla.redhat.com/show_bug.cgi?id=858415 https://bugzilla.redhat.com/show_bug.cgi?id=829657 Regards ___ Gluster-users mailing list Gluster-users@gluster.org http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
Re: [Gluster-users] XFS and MD RAID
* Brian Candler [2012 08 29, 08:48]: > In a test setup (Ubuntu 12.04, gluster 3.3.0, 24 x SATA HD on LSI Megaraid > controllers, MD RAID) I can cause XFS corruption just by throwing some > bonnie++ load at the array - locally without gluster. Randomly found on Google: http://www.jive.nl/nexpres/doku.php?id=nexpres:nexpres_wp8#tests_on_xfs_file_system "It is our opinion that the normalization of XFS behavior on a 24 disks array is due to some proprietary round-robin algorithm on the raid card that caused during the tests on a 12 disks array a 'missing disk' signal that slowed down the pace, even though some downfalls on the 24 disks array still happen every 18/20 files written. We ought to say that the downfall pattern is not related to time delays or file sizes, but it is instead a peculiarity of the XFS file system." Now I'd _really_ like to know if you are using a Megaraid or, as you say at the end, a mpt2sas controller/driver, because I am going to setup a new gluster volume with them, and considering this issue and the ext4 one I don't really know what to choose... Regards ___ Gluster-users mailing list Gluster-users@gluster.org http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
Re: [Gluster-users] [Gluster-devel] FeedBack Requested : Changes to CLI output of 'peer status'
* Stephan von Krawczynski [2012 08 28, 15:31]: > You are not seriously talking about 80 char terminals for an output > that is commonly used by scripts and stuff like nagios, are you? > > > - > > Hostname: 10.70.36.7 > > Uuid: c7283ee7-0e8d-4cb8-8552-a63ab05deaa7 > > State: Peer in Cluster (Connected) > > - Given the '80 char terminals' and the plan to add more fields, I would suggest a kind of key-value output which should be as easy to parse: - cut here - c7283ee7-0e8d-4cb8-8552-a63ab05deaa7: Hostname: 10.70.36.7 c7283ee7-0e8d-4cb8-8552-a63ab05deaa7: State: Connected - cut here - This way you should be able to line-filter the output as if it was all in one line. ___ Gluster-users mailing list Gluster-users@gluster.org http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
Re: [Gluster-users] Failed to get names of volumes
* s19n [2012 08 27, 17:56]: > Additional update: seems that the command (and many others) is not > working because of a 'local lock' being held. I think I am hitting the following, so I'll be looking forward the fix: https://bugzilla.redhat.com/show_bug.cgi?id=843003 ___ Gluster-users mailing list Gluster-users@gluster.org http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
Re: [Gluster-users] Failed to get names of volumes
* s19n [2012 08 27, 11:25]: > - cut here - > # gluster volume status all > operation failed > > Failed to get names of volumes > - cut here - Additional update: seems that the command (and many others) is not working because of a 'local lock' being held. I have tried running the command on the host holding the lock as well as on other hosts, with no success. etc-glusterfs-glusterd.vol.log extract follows: - cut here - [2012-08-27 17:41:17.842441] I [glusterd-volume-ops.c:583:glusterd_handle_cli_statedump_volume] 0-glusterd: Received statedump request for volume storage with options [2012-08-27 17:41:17.842510] E [glusterd-utils.c:277:glusterd_lock] 0-glusterd: Unable to get lock for uuid: 34b665ea-d315-489b-bd0f-172bb6b85ee1, lock held by: 34b665ea-d315-489b-bd0f-172bb6b85ee1 [2012-08-27 17:41:17.842527] E [glusterd-handler.c:453:glusterd_op_txn_begin] 0-management: Unable to acquire local lock, ret: -1 - cut here - Could you please describe why these locks are being held and, if possible, how to clear them? Thank you very much for your kind attention, Best regards ___ Gluster-users mailing list Gluster-users@gluster.org http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
Re: [Gluster-users] Failed to get names of volumes
* Tao Lin [2012 08 27, 17:54]: > There are issues with gluster on ext4,you have to use other file > systems(eg. xfs, ext3) instead of ext4. If you are referring to http://joejulian.name/blog/glusterfs-bit-by-ext4-structure-change/ I don't think I was experienceing that problem, since I shouldn't have an affected kernel version. Update: now I am able to issue the command, and indeed noticed that one of the nodes was offline. In the brick log I found: - cut here - patchset: git://git.gluster.com/glusterfs.git signal received: 7 time of crash: 2012-08-24 23:22:30 configuration details: argp 1 backtrace 1 dlfcn 1 fdatasync 1 libpthread 1 llistxattr 1 setfsid 1 spinlock 1 epoll.h 1 xattr.h 1 st_atim.tv_nsec 1 package-string: glusterfs 3.3.0 /lib/libc.so.6(+0x33af0)[0x7f024da35af0] /usr/lib/libglusterfs.so.0(__dentry_grep+0x8e)[0x7f024e7879de] /usr/lib/libglusterfs.so.0(inode_grep+0x66)[0x7f024e787c56] /usr/lib/glusterfs/3.3.0/xlator/protocol/server.so(resolve_entry_simple+0x91)[0x7f02491eb641] /usr/lib/glusterfs/3.3.0/xlator/protocol/server.so(server_resolve_entry+0x24)[0x7f02491ebd14] /usr/lib/glusterfs/3.3.0/xlator/protocol/server.so(server_resolve+0x98)[0x7f02491ebb88] /usr/lib/glusterfs/3.3.0/xlator/protocol/server.so(server_resolve_all+0x9e)[0x7f02491ebcbe] /usr/lib/glusterfs/3.3.0/xlator/protocol/server.so(resolve_and_resume+0x14)[0x7f02491ebd84] /usr/lib/glusterfs/3.3.0/xlator/protocol/server.so(server_lookup+0x18f)[0x7f024920525f] /usr/lib/libgfrpc.so.0(rpcsvc_handle_rpc_call+0x293)[0x7f024e550ce3] /usr/lib/libgfrpc.so.0(rpcsvc_notify+0x93)[0x7f024e550e53] /usr/lib/libgfrpc.so.0(rpc_transport_notify+0x28)[0x7f024e5518b8] /usr/lib/glusterfs/3.3.0/rpc-transport/socket.so(socket_event_poll_in+0x34)[0x7f024acd0734] /usr/lib/glusterfs/3.3.0/rpc-transport/socket.so(socket_event_handler+0xc7)[0x7f024acd0817] /usr/lib/libglusterfs.so.0(+0x3e394)[0x7f024e79b394] /usr/sbin/glusterfsd(main+0x58a)[0x407aaa] /lib/libc.so.6(__libc_start_main+0xfd)[0x7f024da20c4d] /usr/sbin/glusterfsd[0x404a59] - cut here - Regards ___ Gluster-users mailing list Gluster-users@gluster.org http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
[Gluster-users] Failed to get names of volumes
Hello, Gluster 3.3.0 distributed replicated, ext4 bricks. Since this morning I am unable to check the status of the filesystem: - cut here - # gluster volume status all operation failed Failed to get names of volumes - cut here - Extract from cli.log: - cut here - [2012-08-27 11:10:00.341089] W [rpc-transport.c:174:rpc_transport_load] 0-rpc-transport: missing 'option transport-type'. defaulting to "socket" [2012-08-27 11:10:00.439743] E [cli-rpc-ops.c:5657:gf_cli_status_volume_all] 0-cli: status all failed [2012-08-27 11:10:00.439791] I [input.c:46:cli_batch] 0-: Exiting with: -22 - cut here - "gluster peer" reports all nodes connected but it seems to be known as not really reliable. Any suggestion will be really appreciated. Best regards ___ Gluster-users mailing list Gluster-users@gluster.org http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
Re: [Gluster-users] files on gluster brick that have '---------T' designation.
* Shishir Gowda [2012 08 20, 22:09]: > These are valid files in glusterfs-dht xlator configured volumes. > These are known as link files, which dht uses to maintain files on > the hashed subvol, when the actual data resides in non hashed > subvolumes(rename can lead to these). Hello, since I also have noticed those files, could you please clarify (or point me to the proper documentation) what 'hashed subvolumes' and 'non hashed subvolumes' are? > The cleanup of these files will be taken care of by running > rebalance. When you say rebalance you mean a fix-layout or a full rebalance with data migration? Should I plan to issue a rebalance request periodically to get rid of those files? Thank you very much in advance for your attention, Best regards ___ Gluster-users mailing list Gluster-users@gluster.org http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
[Gluster-users] gfid mismatch on 3.3.0
Hello, I am seeing the following on a gluster 3.3.0 setup (5x2 bricks distributed replicated): - cut here - [2012-08-17 07:52:21.011448] W [dht-common.c:416:dht_lookup_dir_cbk] 0-storage-dht: /index/it/201208/20120812/data: gfid different on storage-replicate-0 - cut here - Is it possible that the problem addressed at the page http://community.gluster.org/a/alert-glusterfs-release-for-gfid-mismatch/ is still appearing with 3.3.0? This is a new setup, not an upgraded one. Thank you very much in advance for your attention, Best regards ___ Gluster-users mailing list Gluster-users@gluster.org http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
[Gluster-users] Replica with non-uniform bricks
Hello, as per http://gluster.org/pipermail/gluster-users/2011-May/007788.html I am advised to use bricks which are uniform in size, but for a number of reasons this is not possible at the moment in my current setup. So the question is: what is the expected behaviour when two bricks with different sizes are coupled to form a replica? Will the larger brick keep accepting writes even after the smallest brick has been filled up? In this scenario, what conditions become relevant in the process of rebalancing a volume? Is there a precedence of the first brick in a replica (as per volume definition)? Thank you very much in advance for your attention, Best regards ___ Gluster-users mailing list Gluster-users@gluster.org http://gluster.org/cgi-bin/mailman/listinfo/gluster-users