Re: [Gluster-users] GlusterD uses 50% of RAM
Hi, today the issue happened once again. Glusterd process was using 80% of RAM and its log was filling up the /var/log. One month ago, when the issue last happened, you suggested to install a patch, so I did this: git fetch git://review.gluster.org/glusterfs refs/changes/28/9328/4 git fetch http://review.gluster.org/glusterfs refs/changes/28/9328/4 git checkout FETCH_HEAD Is this enough to install the patch or I missed something? Thank you Alessandro -Original Message- From: RASTELLI Alessandro Sent: martedì 24 febbraio 2015 10:28 To: 'Atin Mukherjee' Cc: gluster-users@gluster.org Subject: RE: [Gluster-users] GlusterD uses 50% of RAM Hi Atin, I managed to install the patch, it fixed the issue Thank you A. -Original Message- From: Atin Mukherjee [mailto:amukh...@redhat.com] Sent: martedì 24 febbraio 2015 08:03 To: RASTELLI Alessandro Cc: gluster-users@gluster.org Subject: Re: [Gluster-users] GlusterD uses 50% of RAM On 02/20/2015 07:20 PM, RASTELLI Alessandro wrote: I get this: [root@gluster03-mi glusterfs]# git fetch git://review.gluster.org/glusterfs refs/changes/28/9328/4 git checkout FETCH_HEAD fatal: Couldn't find remote ref refs/changes/28/9328/4 What's wrong with that? Is your current branch at 3.6 ? A. -Original Message- From: Atin Mukherjee [mailto:amukh...@redhat.com] Sent: venerdì 20 febbraio 2015 12:54 To: RASTELLI Alessandro Cc: gluster-users@gluster.org Subject: Re: [Gluster-users] GlusterD uses 50% of RAM From the cmd log history I could see lots of volume status commands were triggered parallely. This is a known issue for 3.6 and it would cause a memory leak. http://review.gluster.org/#/c/9328/ should solve it. ~Atin On 02/20/2015 04:36 PM, RASTELLI Alessandro wrote: 10MB log sorry :) -Original Message- From: Atin Mukherjee [mailto:amukh...@redhat.com] Sent: venerdì 20 febbraio 2015 10:49 To: RASTELLI Alessandro; gluster-users@gluster.org Subject: Re: [Gluster-users] GlusterD uses 50% of RAM Could you please share the cmd_history.log glusterd log file to analyze this high memory usage. ~Atin On 02/20/2015 03:10 PM, RASTELLI Alessandro wrote: Hi, I've noticed that one of our 6 gluster 3.6.2 nodes has glusterd process using 50% of RAM, on the other nodes usage is about 5% This can be a bug? Should I restart glusterd daemon? Thank you A From: Volnei Puttini [mailto:vol...@vcplinux.com.br] Sent: lunedì 9 febbraio 2015 18:06 To: RASTELLI Alessandro; gluster-users@gluster.org Subject: Re: [Gluster-users] cannot access to CIFS export Hi Alessandro, My system: CentOS 7 samba-vfs-glusterfs-4.1.1-37.el7_0.x86_64 samba-winbind-4.1.1-37.el7_0.x86_64 samba-libs-4.1.1-37.el7_0.x86_64 samba-common-4.1.1-37.el7_0.x86_64 samba-winbind-modules-4.1.1-37.el7_0.x86_64 samba-winbind-clients-4.1.1-37.el7_0.x86_64 samba-4.1.1-37.el7_0.x86_64 samba-client-4.1.1-37.el7_0.x86_64 glusterfs 3.6.2 built on Jan 22 2015 12:59:57 Try this, work fine for me: [GFSVOL] browseable = No comment = Gluster share of volume gfsvol path = / read only = No guest ok = Yes kernel share modes = No posix locking = No vfs objects = glusterfs glusterfs:loglevel = 7 glusterfs:logfile = /var/log/samba/glusterfs-gfstest.log glusterfs:volume = vgtest glusterfs:volfile_server = 192.168.2.21 On 09-02-2015 14:45, RASTELLI Alessandro wrote: Hi, I've created and started a new replica volume downloadstat with CIFS export enabled on GlusterFS 3.6.2. I can see the following piece has been added automatically to smb.conf: [gluster-downloadstat] comment = For samba share of volume downloadstat vfs objects = glusterfs glusterfs:volume = downloadstat glusterfs:logfile = /var/log/samba/glusterfs-downloadstat.%M.log glusterfs:loglevel = 7 path = / read only = no guest ok = yes I restarted smb service, without errors. When I try to access from Win7 client to \\gluster01-mi\gluster-downloadstatfile:///\\gluster01-mi\gluster-downloadstat it asks me a login (which user do I need to put?) and then gives me error The network path was not found and on Gluster smb.log I see: [2015/02/09 17:21:13.111639, 0] smbd/vfs.c:173(vfs_init_custom) error probing vfs module 'glusterfs': NT_STATUS_UNSUCCESSFUL [2015/02/09 17:21:13.111709, 0] smbd/vfs.c:315(smbd_vfs_init) smbd_vfs_init: vfs_init_custom failed for glusterfs [2015/02/09 17:21:13.111741, 0] smbd/service.c:902(make_connection_snum) vfs_init failed for service gluster-downloadstat Can you explain how to fix? Thanks Alessandro From: gluster-users-boun...@gluster.orgmailto:gluster-users-bounces@glust e r .org [mailto:gluster-users-boun...@gluster.org] On Behalf Of David F. Robinson Sent: domenica 8 febbraio 2015 18:19 To: Gluster Devel; gluster-users@gluster.orgmailto:gluster-users@gluster.org Subject: [Gluster-users] cannot delete non-empty directory I am seeing these
Re: [Gluster-users] What should I do to improve performance ?
On 24-3-2015 12:59, marianna cattani wrote: Hello Ben , whole infrastucture runs on 4 servers with SAS drives 7,200 rpm and a raid controller LSI. Probably the network is oversized compared to the disks and controllers . To verify that libgfapi is operating, is enough that my vm's disks have named as / dev / vd * ? do a 'virsh -r dumpxml yourvm' and have a look at the output should like like: disk type='network' device='disk' snapshot='no' driver name='qemu' type='raw' cache='none' error_policy='stop' io='threads'/ source protocol='gluster' name='GlusterSSD/e11866bf-1120-4c87-a992-9be32f110b8d/images/09ca5046-ab7a-11d8-9f2a-ffac3e89a6ad/042a9f7d-15be-443d-bcf7-a2ba8db43f72' host name='test.com' port='0'/ /source target dev='vda' bus='virtio'/ serial09ca5046-ab7a-4bd8-9f2a-ffac3e89a6ad/serial boot order='1'/ alias name='virtio-disk0'/ address type='pci' domain='0x' bus='0x00' slot='0x04' function='0x0'/ /disk The first line is import type=network indicates that its using libgfapi. Joop ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] What should I do to improve performance ?
Openstack doesn't have vsdm, it should be a configuration option: qemu_allowed_storage_drivers=gluster But, however , the machine is generated with the xml that you see. Now I try to write to the OpenStack's mailing list . tnx M 2015-03-24 15:14 GMT+01:00 noc n...@nieuwland.nl: On 24-3-2015 14:39, marianna cattani wrote: Many many thanks Mine is different :'( root@nodo-4:~# virsh -r dumpxml instance-002c | grep disk disk type='file' device='disk' source file='/var/lib/nova/instances/ef84920d-2009-42a2-90de-29d9bd5e8512/disk'/ alias name='virtio-disk0'/ /disk Thats a fuse connection. I'm running ovirt + glusterfs where vdsm is a special version with libgfapi support. Don't know if Openstack has that too? Joop ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] What should I do to improve performance ?
Many many thanks Mine is different :'( root@nodo-4:~# virsh -r dumpxml instance-002c | grep disk disk type='file' device='disk' source file='/var/lib/nova/instances/ef84920d-2009-42a2-90de-29d9bd5e8512/disk'/ alias name='virtio-disk0'/ /disk Now I investigate more. BR. M. 2015-03-24 14:29 GMT+01:00 noc n...@nieuwland.nl: On 24-3-2015 12:59, marianna cattani wrote: Hello Ben , whole infrastucture runs on 4 servers with SAS drives 7,200 rpm and a raid controller LSI. Probably the network is oversized compared to the disks and controllers . To verify that libgfapi is operating, is enough that my vm's disks have named as / dev / vd * ? do a 'virsh -r dumpxml yourvm' and have a look at the output should like like: disk type='network' device='disk' snapshot='no' driver name='qemu' type='raw' cache='none' error_policy='stop' io='threads'/ source protocol='gluster' name='GlusterSSD/e11866bf-1120-4c87-a992-9be32f110b8d/images/09ca5046-ab7a-11d8-9f2a-ffac3e89a6ad/042a9f7d-15be-443d-bcf7-a2ba8db43f72' host name='test.com' port='0'/ /source target dev='vda' bus='virtio'/ serial09ca5046-ab7a-4bd8-9f2a-ffac3e89a6ad/serial boot order='1'/ alias name='virtio-disk0'/ address type='pci' domain='0x' bus='0x00' slot='0x04' function='0x0'/ /disk The first line is import type=network indicates that its using libgfapi. Joop ___ 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] lots of nfs.log activity since upgrading to 3.4.6
Hello list, I have a few Wordpress sites served via NFS on Gluster. Since upgrading 3.4.6, I'm seeing a non-trivial amount (1-2 million depending on how busy the blogs are, about 4 gigs in the last three weeks) entries like this appear in nfs.log: [2015-03-24 13:49:17.314003] I [dht-common.c:1000:dht_lookup_everywhere_done] 0-wpblog-storage-dht: STATUS: hashed_subvol wpblog-storage-replicate-0 cached_subvol null [2015-03-24 13:49:17.355722] I [dht-common.c:1000:dht_lookup_everywhere_done] 0-wpblog-storage-dht: STATUS: hashed_subvol wpblog-storage-replicate-0 cached_subvol null [2015-03-24 13:49:17.616073] I [dht-common.c:1000:dht_lookup_everywhere_done] 0-wpblog-storage-dht: STATUS: hashed_subvol wpblog-storage-replicate-0 cached_subvol null It doesn't seem to be a big problem, it's just an INFO log, but it definitely wasn't there with 3.4.5. Can anyone give me any insight into what's going on? -Matt ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] What should I do to improve performance ?
On 24-3-2015 14:39, marianna cattani wrote: Many many thanks Mine is different :'( root@nodo-4:~# virsh -r dumpxml instance-002c | grep disk disk type='file' device='disk' source file='/var/lib/nova/instances/ef84920d-2009-42a2-90de-29d9bd5e8512/disk'/ alias name='virtio-disk0'/ /disk Thats a fuse connection. I'm running ovirt + glusterfs where vdsm is a special version with libgfapi support. Don't know if Openstack has that too? Joop ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] What should I do to improve performance ?
Hello Ben , whole infrastucture runs on 4 servers with SAS drives 7,200 rpm and a raid controller LSI. Probably the network is oversized compared to the disks and controllers . To verify that libgfapi is operating, is enough that my vm's disks have named as / dev / vd * ? I'm not sure that I understand the test which you suggest , if this is my mount point /var/lib/nova/instances I could run: bonnie++ -d /var/lib/nova/instances -r 2048 then on the vm that use /var/lib/nova/instances i run the same thing ? Is it right ? BR M. 2015-03-23 19:57 GMT+01:00 Ben Turner btur...@redhat.com: - Original Message - From: marianna cattani marianna.catt...@gmail.com To: gluster-users@gluster.org Sent: Monday, March 23, 2015 6:09:41 AM Subject: [Gluster-users] What should I do to improve performance ? Dear all, I followed the tutorial I read at this link : http://www.gluster.org/documentation/use_cases/Virt-store-usecase/ I have 4 nodes configured as a linked list , each node also performs virtual machines with KVM and mounts on its ip address, like this: 172.16.155.12:/nova /var/lib/nova/instances glusterfs defaults,_netdev 0 0 Each node has two nic (ten giga) bonded in mode 4. What can I do to further improve the speed ? What kind of disks are back ending your 10G NICs? Are you using FUSE or libgfapi to connect to gluster from your hypervisor? What kind of speeds are you expecting vs seeing in your environment? We need to understand what your HW can do first then gather some data running on gluster and compare the two. As a rule of thumb with replica 2 you should see about: throughput = ( NIC line speed / 2 ) - 20% overhead As long as your disks can service it. If you are seeing about that on the gluster mounts then go inside one of the VMs and run the same test, the VM should get something similar. If you aren't seeing at least 400 MB / sec on sequential writes and 500-700 MB /sec on reads then there may be something off in your storage stack. -b BR. M. ___ 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
Re: [Gluster-users] Out of memory Gluster 3.6.2
http://review.gluster.org/9269 is going to solve this problem, it will be available in the coming 3.6 z stream release. You can refer to the patch to understand the issue and the solution. Let me know in case of any clarification required. ~Atin On 03/24/2015 02:44 PM, Atin Mukherjee wrote: From the cmd_log_history it looks like there are frequent gluster profile commands which were executed that may have lead to OOM issue. I will analyze it further and get back to you. ~Atin On 03/24/2015 02:20 PM, Félix de Lelelis wrote: Hi, Only a monitoring system, Definitely, I canalize all checks by one script that lock all other monitoring checks and only there is one process check gluster. I send you cmd_log_history of the 2 nodes. Thanks. 2015-03-24 9:18 GMT+01:00 Atin Mukherjee amukh...@redhat.com: Could you tell us what activities were run in the cluster? cmd_log_history across all the nodes would give a clear picture of it. ~Atin On 03/24/2015 01:03 PM, Félix de Lelelis wrote: Hi, Today, Glusterd daemon has been killed due to excessive memory consumption: [3505254.762715] Out of memory: Kill process 7780 (glusterd) score 581 or sacrifice child [3505254.763451] Killed process 7780 (glusterd) total-vm:3537640kB, anon-rss:1205240kB, file-rss:672kB I have installed gluster 3.6.2 on centos 7. There in any way to avoid this whithout the need to kill the process?Simply need more memory? Thanks. ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users -- ~Atin -- ~Atin ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] gluster peer probe error (v3.6.2)
Sure I am. Unfortunately it didn't change the result... # killall glusterd # ps -ef | grep gluster root 15755 657 0 18:35 ttyS000:00:00 grep gluster # rm /var/lib/glusterd/peers/* # /usr/sbin/glusterd -p /var/run/glusterd.pid # gluster peer probe 10.32.1.144 # (I killed glusterd and removed the files on both servers.) Regards Andreas On 03/24/15 05:36, Atin Mukherjee wrote: If you are okay to do a fresh set up I would recommend you to clean up /var/lib/glusterd/peers/* and then restart glusterd in both the nodes and then try peer probing. ~Atin On 03/23/2015 06:44 PM, Andreas wrote: Hi, # gluster peer detach 10.32.1.144 (No output here. Similar to the problem with 'gluster peer probe'.) # gluster peer detach 10.32.1.144 force peer detach: failed: Peer is already being detached from cluster. Check peer status by running gluster peer status # gluster peer status Number of Peers: 1 Hostname: 10.32.1.144 Uuid: 82cdb873-28cc-4ed0-8cfe-2b6275770429 State: Probe Sent to Peer (Connected) # ping 10.32.1.144 PING 10.32.1.144 (10.32.1.144): 56 data bytes 64 bytes from 10.32.1.144: seq=0 ttl=64 time=1.811 ms 64 bytes from 10.32.1.144: seq=1 ttl=64 time=1.834 ms ^C --- 10.32.1.144 ping statistics --- 2 packets transmitted, 2 packets received, 0% packet loss round-trip min/avg/max = 1.811/1.822/1.834 ms As previously stated, this problem seems to be similar to what I experienced with 'gluster peer probe'. I can reboot the server, but the situation will be the same (I've tried this many times). Any ideas of which ports to investigate and how to do it to get the most reliable result? Anything else that could cause this? Regards Andreas On 03/23/15 11:10, Atin Mukherjee wrote: On 03/23/2015 03:28 PM, Andreas Hollaus wrote: 2Hi, This network problem is persistent. However, I can ping the server so guess it depends on the port no, right? I tried to telnet to port 24007, but I was not sure how to interpret the result as I got no respons and no timeout (it just seemed to be waiting for something). That's why I decided to install nmap, but according to that tool the port was accessible. Are there any other ports that are vital to gluster peer probe? When you say 'deprobe', I guess you mean 'gluster peer detach'? That command shows similar behaviour to gluster peer probe. Yes I meant peer detach. How about gluster peer detach force? Regards Andreas On 03/23/15 05:34, Atin Mukherjee wrote: On 03/22/2015 07:11 PM, Andreas Hollaus wrote: Hi, I hope that these are the logs that you requested. Logs from 10.32.0.48: -- # more /var/log/glusterfs/.cmd_log_history [2015-03-19 13:52:03.277438] : peer probe 10.32.1.144 : FAILED : Probe returned with unknown errno -1 # more /var/log/glusterfs/etc-glusterfs-glusterd.vol.log [2015-03-19 13:41:31.241768] I [MSGID: 100030] [glusterfsd.c:2018:main] 0-/usr/s bin/glusterd: Started running /usr/sbin/glusterd version 3.6.2 (args: /usr/sbin/ glusterd -p /var/run/glusterd.pid) [2015-03-19 13:41:31.245352] I [glusterd.c:1214:init] 0-management: Maximum allo wed open file descriptors set to 65536 [2015-03-19 13:41:31.245432] I [glusterd.c:1259:init] 0-management: Using /var/l ib/glusterd as working directory [2015-03-19 13:41:31.247826] I [glusterd-store.c:2063:glusterd_restore_op_versio n] 0-management: Detected new install. Setting op-version to maximum : 30600 [2015-03-19 13:41:31.247902] I [glusterd-store.c:3497:glusterd_store_retrieve_mi ssed_snaps_list] 0-management: No missed snaps list. Final graph: +--+ 1: volume management 2: type mgmt/glusterd 3: option rpc-auth.auth-glusterfs on 4: option rpc-auth.auth-unix on 5: option rpc-auth.auth-null on 6: option transport.socket.listen-backlog 128 7: option ping-timeout 30 8: option transport.socket.read-fail-log off 9: option transport.socket.keepalive-interval 2 10: option transport.socket.keepalive-time 10 11: option transport-type socket 12: option working-directory /var/lib/glusterd 13: end-volume 14: +--+ [2015-03-19 13:42:02.258403] I [glusterd-handler.c:1015:__glusterd_handle_cli_pr obe] 0-glusterd: Received CLI probe req 10.32.1.144 24007 [2015-03-19 13:42:02.259456] I [glusterd-handler.c:3165:glusterd_probe_begin] 0- glusterd: Unable to find peerinfo for host: 10.32.1.144 (24007) [2015-03-19 13:42:02.259664] I [rpc-clnt.c:969:rpc_clnt_connection_init] 0-manag ement: setting frame-timeout to 600 [2015-03-19 13:42:02.260488] I [glusterd-handler.c:3098:glusterd_friend_add] 0-m anagement: connect returned 0 [2015-03-19 13:42:02.270316] I [glusterd.c:176:glusterd_uuid_generate_save] 0-ma nagement: generated UUID:
Re: [Gluster-users] [Gluster-devel] Revamping the GlusterFS Documentation...
Yes, We are thinking about a front/landing page for the documentation with better UI/UX. The minor changes that you had suggested, have been made :) -Shravan - Original Message - From: Justin Clift jus...@gluster.org To: Shravan Chandrashekar schan...@redhat.com Cc: gluster-users@gluster.org, gluster-de...@gluster.org, Anjana Suparna Sriram asri...@redhat.com, s...@redhat.com Sent: Tuesday, March 24, 2015 3:04:39 AM Subject: Re: [Gluster-devel] Revamping the GlusterFS Documentation... On 23 Mar 2015, at 07:01, Shravan Chandrashekar schan...@redhat.com wrote: Hi All, The Gluster Filesystem documentation is not user friendly and fragmented and this has been the feedback we have been receiving. We got back to our drawing board and blueprints and realized that the content was scattered at various places. These include: [Static HTML] http://www.gluster.org/documentation/ [Mediawiki] http://www.gluster.org/community/documentation/ [In-source] https://github.com/gluster/glusterfs/tree/master/doc [Markdown] https://github.com/GlusterFS/Notes and so on… Hence, we started by curating content from various sources including gluster.org static HTML documentation, glusterfs github repository, various blog posts and the Community wiki. We also felt the need to improve the community member's experience with Gluster documentation. This led us to put some thought into the user interface. As a result we came up with a page which links all content into a single landing page: http://www.gluster.org/community/documentation/index.php/Staged_Docs This is just our first step to improve our community docs and enhance the community contribution towards documentation. I would like to thank Humble Chirammal and Anjana Sriram for the suggestions and directions during the entire process. I am sure there is lot of scope for improvement. Hence, request you all to review the content and provide your suggestions. Looks like a good effort. Is the general concept for this to become the front/landing page for the main wiki? Also some initial thoughts: * Gluster Ant Logo image - The first letter REALLY looks like a C (to me), not a G. Reads as Cluster for me... That aside, it looks really good. :) * Getting Started section ... move it up maybe, before the Terminology / Architecture / Additional Resources bit This is to make it more obvious for new people. * Terminologies should probably be Terminology, as Terminology is kind of both singular and plural. * All that Developers need to know → Everything Developers need to know They're my first thoughts anyway. :) + Justin -- GlusterFS - http://www.gluster.org An open source, distributed file system scaling to several petabytes, and handling thousands of clients. My personal twitter: twitter.com/realjustinclift ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Revamping the GlusterFS Documentation...
Hi Shravan, Having recently set up geo-replication of Gluster v3.5.3 I can tell you that there is effectively almost no documentation. The documentation that does exists is primarily focused on describing the differences between the current and previous versions. That's completely useless to someone wanting to set it up for the first time and not a whole lot better for someone who has upgraded. The first, and perhaps most crucial, piece of information that's missing is installation requirements. Nowhere have I been able to find out exactly which components are required on either the master or the slave. In my case this was determined by pure trial and error. i.e. Install what I think should be needed and then continue installing components until it starts to work. Even once that part is done, there is a LOT of documentation missing. I recall that when I first set up geo-replication (v3.2 or v3.3?) I was able to follow clear and simple step by step instructions that almost guaranteed success. regards, John On 23/03/15 18:01, Shravan Chandrashekar wrote: *Hi All, * *The Gluster Filesystem documentation is not user friendly and fragmented and this has been the feedback we have been receiving. * *We got back to our drawing board and blueprints and realized that the content was scattered at various places. These include: * *[Static HTML] http://www.gluster.org/documentation/ * *[Mediawiki] http://www.gluster.org/community/documentation/ * *[In-source] https://github.com/gluster/glusterfs/tree/master/doc * *[Markdown] https://github.com/GlusterFS/Notes * *and so on… * *Hence, we started by curating content from various sources including gluster.org static HTML documentation, glusterfs github repository, * *various blog posts and the Community wiki. We also felt the need to improve the community member's experience with Gluster documentation. This led us to put some thought into the user interface. As a result we came up with a page which links all content into a single landing page: * *http://www.gluster.org/community/documentation/index.php/Staged_Docs * *This is just our first step to improve our community docs and enhance the community contribution towards documentation. I would like to thank Humble Chirammal and Anjana Sriram for the suggestions and directions during the entire process. I am sure there is lot of scope for improvement. * *Hence, request you all to review the content and provide your suggestions. * Regards, Shravan Chandra __ This email has been scanned by the Symantec Email Security.cloud service. For more information please visit http://www.symanteccloud.com __ ___ 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
Re: [Gluster-users] gluster peer probe error (v3.6.2)
Hi, The problem seems to be solved now(!). We discovered that the global options file (/var/lib/glusterd/options) generated an error in the log file: [1970-01-01 00:01:24.423024] E [glusterd-utils.c:5760:glusterd_compare_friend_da ta] 0-management: Importing global options failed For some reason the file was missing previously, but that didn't cause any major problems to glusterd (except for an error message about the missing file). However, when an empty options file was created to get rid of that error message, this new message appeared which seems to be more serious than the previous. When the contents 'global-option-version=0' was added to that options file, all those error messages disappeared and 'gluster peer probe' started to work as expected again. Not that obvious, at least not for me. Anyway, thanks for your efforts in trying to solve this problem. Regards Andreas On 03/24/15 10:32, Andreas wrote: Sure I am. Unfortunately it didn't change the result... # killall glusterd # ps -ef | grep gluster root 15755 657 0 18:35 ttyS000:00:00 grep gluster # rm /var/lib/glusterd/peers/* # /usr/sbin/glusterd -p /var/run/glusterd.pid # gluster peer probe 10.32.1.144 # (I killed glusterd and removed the files on both servers.) Regards Andreas On 03/24/15 05:36, Atin Mukherjee wrote: If you are okay to do a fresh set up I would recommend you to clean up /var/lib/glusterd/peers/* and then restart glusterd in both the nodes and then try peer probing. ~Atin On 03/23/2015 06:44 PM, Andreas wrote: Hi, # gluster peer detach 10.32.1.144 (No output here. Similar to the problem with 'gluster peer probe'.) # gluster peer detach 10.32.1.144 force peer detach: failed: Peer is already being detached from cluster. Check peer status by running gluster peer status # gluster peer status Number of Peers: 1 Hostname: 10.32.1.144 Uuid: 82cdb873-28cc-4ed0-8cfe-2b6275770429 State: Probe Sent to Peer (Connected) # ping 10.32.1.144 PING 10.32.1.144 (10.32.1.144): 56 data bytes 64 bytes from 10.32.1.144: seq=0 ttl=64 time=1.811 ms 64 bytes from 10.32.1.144: seq=1 ttl=64 time=1.834 ms ^C --- 10.32.1.144 ping statistics --- 2 packets transmitted, 2 packets received, 0% packet loss round-trip min/avg/max = 1.811/1.822/1.834 ms As previously stated, this problem seems to be similar to what I experienced with 'gluster peer probe'. I can reboot the server, but the situation will be the same (I've tried this many times). Any ideas of which ports to investigate and how to do it to get the most reliable result? Anything else that could cause this? Regards Andreas On 03/23/15 11:10, Atin Mukherjee wrote: On 03/23/2015 03:28 PM, Andreas Hollaus wrote: 2Hi, This network problem is persistent. However, I can ping the server so guess it depends on the port no, right? I tried to telnet to port 24007, but I was not sure how to interpret the result as I got no respons and no timeout (it just seemed to be waiting for something). That's why I decided to install nmap, but according to that tool the port was accessible. Are there any other ports that are vital to gluster peer probe? When you say 'deprobe', I guess you mean 'gluster peer detach'? That command shows similar behaviour to gluster peer probe. Yes I meant peer detach. How about gluster peer detach force? Regards Andreas On 03/23/15 05:34, Atin Mukherjee wrote: On 03/22/2015 07:11 PM, Andreas Hollaus wrote: Hi, I hope that these are the logs that you requested. Logs from 10.32.0.48: -- # more /var/log/glusterfs/.cmd_log_history [2015-03-19 13:52:03.277438] : peer probe 10.32.1.144 : FAILED : Probe returned with unknown errno -1 # more /var/log/glusterfs/etc-glusterfs-glusterd.vol.log [2015-03-19 13:41:31.241768] I [MSGID: 100030] [glusterfsd.c:2018:main] 0-/usr/s bin/glusterd: Started running /usr/sbin/glusterd version 3.6.2 (args: /usr/sbin/ glusterd -p /var/run/glusterd.pid) [2015-03-19 13:41:31.245352] I [glusterd.c:1214:init] 0-management: Maximum allo wed open file descriptors set to 65536 [2015-03-19 13:41:31.245432] I [glusterd.c:1259:init] 0-management: Using /var/l ib/glusterd as working directory [2015-03-19 13:41:31.247826] I [glusterd-store.c:2063:glusterd_restore_op_versio n] 0-management: Detected new install. Setting op-version to maximum : 30600 [2015-03-19 13:41:31.247902] I [glusterd-store.c:3497:glusterd_store_retrieve_mi ssed_snaps_list] 0-management: No missed snaps list. Final graph: +--+ 1: volume management 2: type mgmt/glusterd 3: option rpc-auth.auth-glusterfs on 4: option rpc-auth.auth-unix on 5: option rpc-auth.auth-null on 6: option transport.socket.listen-backlog 128 7: option ping-timeout 30 8: option transport.socket.read-fail-log off 9:
Re: [Gluster-users] tune2fs exited with non-zero exit status
Ah, that is handy to know. Will this patch get applied to the 3.5. release stream or am I going to have to look at moving onto 3.6 at some point. Thanks Paul -- Paul Osborne Senior Systems Engineer Infrastructure Services IT Department Canterbury Christ Church University -Original Message- From: gluster-users-boun...@gluster.org [mailto:gluster-users-boun...@gluster.org] On Behalf Of Atin Mukherjee Sent: 18 March 2015 16:24 To: Vitaly Lipatov; gluster-users@gluster.org Subject: Re: [Gluster-users] tune2fs exited with non-zero exit status On 03/18/2015 08:04 PM, Vitaly Lipatov wrote: Osborne, Paul (paul.osbo...@canterbury.ac.uk) писал 2015-03-16 19:22: Hi, I am just looking through my logs and am seeing a lot of entries of the form: [2015-03-16 16:02:55.553140] I [glusterd-handler.c:3530:__glusterd_handle_status_volume] 0-management: Received status volume req for volume wiki [2015-03-16 16:02:55.561173] E [glusterd-utils.c:5140:glusterd_add_inode_size_to_dict] 0-management: tune2fs exited with non-zero exit status [2015-03-16 16:02:55.561204] E [glusterd-utils.c:5166:glusterd_add_inode_size_to_dict] 0-management: failed to get inode size Having had a rummage I *SUSPECT* it is because gluster is trying to get the volume status by querying the superblock on the filesystem for a brick volume. However this is an issue as when the volume was created it was done so in the form: I believe it is a bug with tune2fs missed device argument, introduced this patch http://review.gluster.org/#/c/8134/ Could you send a patch to fix this problem? You can refer to [1] for the workflow to send a patch. [1] http://www.gluster.org/community/documentation/index.php/Simplified_dev_workflow ~Atin ___ 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 mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] GlusterD uses 50% of RAM
Can you share the recent cmd_log_history again. Have you triggered lots of volume set commands? In recent past we have discovered that volume set has some potential memory leak. ~Atin On 03/24/2015 07:49 PM, RASTELLI Alessandro wrote: Hi, today the issue happened once again. Glusterd process was using 80% of RAM and its log was filling up the /var/log. One month ago, when the issue last happened, you suggested to install a patch, so I did this: git fetch git://review.gluster.org/glusterfs refs/changes/28/9328/4 git fetch http://review.gluster.org/glusterfs refs/changes/28/9328/4 git checkout FETCH_HEAD Is this enough to install the patch or I missed something? Thank you Alessandro -Original Message- From: RASTELLI Alessandro Sent: martedì 24 febbraio 2015 10:28 To: 'Atin Mukherjee' Cc: gluster-users@gluster.org Subject: RE: [Gluster-users] GlusterD uses 50% of RAM Hi Atin, I managed to install the patch, it fixed the issue Thank you A. -Original Message- From: Atin Mukherjee [mailto:amukh...@redhat.com] Sent: martedì 24 febbraio 2015 08:03 To: RASTELLI Alessandro Cc: gluster-users@gluster.org Subject: Re: [Gluster-users] GlusterD uses 50% of RAM On 02/20/2015 07:20 PM, RASTELLI Alessandro wrote: I get this: [root@gluster03-mi glusterfs]# git fetch git://review.gluster.org/glusterfs refs/changes/28/9328/4 git checkout FETCH_HEAD fatal: Couldn't find remote ref refs/changes/28/9328/4 What's wrong with that? Is your current branch at 3.6 ? A. -Original Message- From: Atin Mukherjee [mailto:amukh...@redhat.com] Sent: venerdì 20 febbraio 2015 12:54 To: RASTELLI Alessandro Cc: gluster-users@gluster.org Subject: Re: [Gluster-users] GlusterD uses 50% of RAM From the cmd log history I could see lots of volume status commands were triggered parallely. This is a known issue for 3.6 and it would cause a memory leak. http://review.gluster.org/#/c/9328/ should solve it. ~Atin On 02/20/2015 04:36 PM, RASTELLI Alessandro wrote: 10MB log sorry :) -Original Message- From: Atin Mukherjee [mailto:amukh...@redhat.com] Sent: venerdì 20 febbraio 2015 10:49 To: RASTELLI Alessandro; gluster-users@gluster.org Subject: Re: [Gluster-users] GlusterD uses 50% of RAM Could you please share the cmd_history.log glusterd log file to analyze this high memory usage. ~Atin On 02/20/2015 03:10 PM, RASTELLI Alessandro wrote: Hi, I've noticed that one of our 6 gluster 3.6.2 nodes has glusterd process using 50% of RAM, on the other nodes usage is about 5% This can be a bug? Should I restart glusterd daemon? Thank you A From: Volnei Puttini [mailto:vol...@vcplinux.com.br] Sent: lunedì 9 febbraio 2015 18:06 To: RASTELLI Alessandro; gluster-users@gluster.org Subject: Re: [Gluster-users] cannot access to CIFS export Hi Alessandro, My system: CentOS 7 samba-vfs-glusterfs-4.1.1-37.el7_0.x86_64 samba-winbind-4.1.1-37.el7_0.x86_64 samba-libs-4.1.1-37.el7_0.x86_64 samba-common-4.1.1-37.el7_0.x86_64 samba-winbind-modules-4.1.1-37.el7_0.x86_64 samba-winbind-clients-4.1.1-37.el7_0.x86_64 samba-4.1.1-37.el7_0.x86_64 samba-client-4.1.1-37.el7_0.x86_64 glusterfs 3.6.2 built on Jan 22 2015 12:59:57 Try this, work fine for me: [GFSVOL] browseable = No comment = Gluster share of volume gfsvol path = / read only = No guest ok = Yes kernel share modes = No posix locking = No vfs objects = glusterfs glusterfs:loglevel = 7 glusterfs:logfile = /var/log/samba/glusterfs-gfstest.log glusterfs:volume = vgtest glusterfs:volfile_server = 192.168.2.21 On 09-02-2015 14:45, RASTELLI Alessandro wrote: Hi, I've created and started a new replica volume downloadstat with CIFS export enabled on GlusterFS 3.6.2. I can see the following piece has been added automatically to smb.conf: [gluster-downloadstat] comment = For samba share of volume downloadstat vfs objects = glusterfs glusterfs:volume = downloadstat glusterfs:logfile = /var/log/samba/glusterfs-downloadstat.%M.log glusterfs:loglevel = 7 path = / read only = no guest ok = yes I restarted smb service, without errors. When I try to access from Win7 client to \\gluster01-mi\gluster-downloadstatfile:///\\gluster01-mi\gluster-downloadstat it asks me a login (which user do I need to put?) and then gives me error The network path was not found and on Gluster smb.log I see: [2015/02/09 17:21:13.111639, 0] smbd/vfs.c:173(vfs_init_custom) error probing vfs module 'glusterfs': NT_STATUS_UNSUCCESSFUL [2015/02/09 17:21:13.111709, 0] smbd/vfs.c:315(smbd_vfs_init) smbd_vfs_init: vfs_init_custom failed for glusterfs [2015/02/09 17:21:13.111741, 0] smbd/service.c:902(make_connection_snum) vfs_init failed for service gluster-downloadstat Can you explain how to fix? Thanks Alessandro From: gluster-users-boun...@gluster.orgmailto:gluster-users-bounces@glust e r .org
Re: [Gluster-users] Out of memory Gluster 3.6.2
From the cmd_log_history it looks like there are frequent gluster profile commands which were executed that may have lead to OOM issue. I will analyze it further and get back to you. ~Atin On 03/24/2015 02:20 PM, Félix de Lelelis wrote: Hi, Only a monitoring system, Definitely, I canalize all checks by one script that lock all other monitoring checks and only there is one process check gluster. I send you cmd_log_history of the 2 nodes. Thanks. 2015-03-24 9:18 GMT+01:00 Atin Mukherjee amukh...@redhat.com: Could you tell us what activities were run in the cluster? cmd_log_history across all the nodes would give a clear picture of it. ~Atin On 03/24/2015 01:03 PM, Félix de Lelelis wrote: Hi, Today, Glusterd daemon has been killed due to excessive memory consumption: [3505254.762715] Out of memory: Kill process 7780 (glusterd) score 581 or sacrifice child [3505254.763451] Killed process 7780 (glusterd) total-vm:3537640kB, anon-rss:1205240kB, file-rss:672kB I have installed gluster 3.6.2 on centos 7. There in any way to avoid this whithout the need to kill the process?Simply need more memory? Thanks. ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users -- ~Atin -- ~Atin ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] [Gluster-devel] Fwd: Change in ffilz/nfs-ganesha[next]: pNFS code drop enablement and checkpatch warnings fixed
On 03/23/2015 12:49 PM, Anand Subramanian wrote: FYI. GlusterFS vols can now be accessed via NFSv4.1 pNFS protocol (mount -t nfs -o minorversion=1 ...) from nfs-ganesha 2.2-rc5 onwards. Note: one fix is to go into libgfapi to fix up using anonymous fd's in ds_write/make_ds_handle() (Avati's sugeestion that really helps here). Once Jiffin or myself get that fix in, a good large file performance can be seen with pNFS vs V4. All thanks and credit to Jiffin for his terrific effort in coding things up quickly and for fixing bugs. Anand Great news! I did a quick check in the docs directory i.e. https://github.com/gluster/glusterfs/tree/master/doc to see if we have any documentation about nfs-ganesha or pNFS and glusterfs integration, but did not find any. I think without howtos around this will hamper the adoption of this feature among users. So if we can get some documentation for this, it will be awesome. Thanks, Lala Forwarded Message Subject: Change in ffilz/nfs-ganesha[next]: pNFS code drop enablement and checkpatch warnings fixed Date: Sat, 21 Mar 2015 01:04:30 +0100 From: GerritHub supp...@gerritforge.com Reply-To: ffilz...@mindspring.com To: Anand Subramanian ana...@redhat.com CC: onnfrhvruutnzhnaq.-g...@noclue.notk.org From Frank Filzffilz...@mindspring.com: Frank Filz has submitted this change and it was merged. Change subject: pNFS code drop enablement and checkpatch warnings fixed .. pNFS code drop enablement and checkpatch warnings fixed Change-Id: Ia8c58dd6d6326f692681f76b96f29c630db21a92 Signed-off-by: Anand Subramanianana...@redhat.com --- A src/FSAL/FSAL_GLUSTER/ds.c M src/FSAL/FSAL_GLUSTER/export.c M src/FSAL/FSAL_GLUSTER/gluster_internal.h M src/FSAL/FSAL_GLUSTER/handle.c M src/FSAL/FSAL_GLUSTER/main.c A src/FSAL/FSAL_GLUSTER/mds.c 6 files changed, 993 insertions(+), 0 deletions(-) -- To view, visithttps://review.gerrithub.io/221683 To unsubscribe, visithttps://review.gerrithub.io/settings Gerrit-MessageType: merged Gerrit-Change-Id: Ia8c58dd6d6326f692681f76b96f29c630db21a92 Gerrit-PatchSet: 1 Gerrit-Project: ffilz/nfs-ganesha Gerrit-Branch: next Gerrit-Owner: Anand Subramanianana...@redhat.com Gerrit-Reviewer: Frank Filzffilz...@mindspring.com Gerrit-Reviewer:onnfrhvruutnzhnaq.-g...@noclue.notk.org ___ Gluster-devel mailing list gluster-de...@gluster.org http://www.gluster.org/mailman/listinfo/gluster-devel ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
[Gluster-users] Do i need to configure vol files?
Dear List. i am very happy with my glusterfs replica 2 so far. i have two volumes (one 16TB raid5 on each server) devided into 2 8 TB partitions which are used as volumes I configured some volume options via gluster volume set the options so far: performance.cache-size: 34359738368 performance.io-thread-count: 32 cluster.data-self-heal-algorithm: full performance.quick-read: on Now i read in the internet many posts about editing vol files. do i need to configure my vol files? i have one volfile under /etc/gluster/glusterd.vol and for each volume another two (for each server) under /var/lib/glusterd/volname/ But in neither volfile i am able to find a translator section of the options that i reconfigured. Do i need to create this sections manually? Background: actually we are driving a 5 node proxmox cluster which uses NFS as storage from a fileserver. Now we want to replace this setup with a 3 node proxmox cluster, from which 2 machines act as gluster replica 2 All test showed that the new setup is 200% faster than the old one, except a linux DB2 vm, which takes double time to write into database (on insert and delete oprations) I want to optimize gluster for use of the databases and will now test the quickread option, do you think this will bring any performance? Thank you Alex ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Access data directly from underlying storage
Thank you once again for your input. It's highly appreciated. On Sat, Mar 21, 2015 at 9:53 AM, Melkor Lord melkor.l...@gmail.com wrote: On Thu, Mar 19, 2015 at 9:11 PM, Rumen Telbizov telbi...@gmail.com wrote: Thank you for your answer Melkor. You're welcome! This is the kind of experience I was looking for actually. I am happy that it has worked fine for you. Anybody coming across any issues while reading directly from the underlying disk? I don't think you'll find any issues, at least on the replica scenario since the copied data is exactly the same for every brick in the cluster. After all, Gluster only exports a directory content accross the network. I don't really see how it could mess things up, especially if all you do is reading the files on the exported directory. I see the workflow basically as NFS or Samba. You can acces the exported root and remove/add files from it without creating havock for the clients except the usual can't access xxx if you remove/rename a file before the client refreshes the directory listing. I won't try such things though (not outside a test environment only) on a distributed scenario since in that particular case, the files are splitted in bits. Still, I think read only is always safe :-) -- Unix _IS_ user friendly, it's just selective about who its friends are. -- Rumen Telbizov Unix Systems Administrator http://telbizov.com ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
[Gluster-users] Out of memory Gluster 3.6.2
Hi, Today, Glusterd daemon has been killed due to excessive memory consumption: [3505254.762715] Out of memory: Kill process 7780 (glusterd) score 581 or sacrifice child [3505254.763451] Killed process 7780 (glusterd) total-vm:3537640kB, anon-rss:1205240kB, file-rss:672kB I have installed gluster 3.6.2 on centos 7. There in any way to avoid this whithout the need to kill the process?Simply need more memory? Thanks. ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] [Gluster-devel] Fwd: Change in ffilz/nfs-ganesha[next]: pNFS code drop enablement and checkpatch warnings fixed
On 24/03/15 12:37, Lalatendu Mohanty wrote: On 03/23/2015 12:49 PM, Anand Subramanian wrote: FYI. GlusterFS vols can now be accessed via NFSv4.1 pNFS protocol (mount -t nfs -o minorversion=1 ...) from nfs-ganesha 2.2-rc5 onwards. Note: one fix is to go into libgfapi to fix up using anonymous fd's in ds_write/make_ds_handle() (Avati's sugeestion that really helps here). Once Jiffin or myself get that fix in, a good large file performance can be seen with pNFS vs V4. All thanks and credit to Jiffin for his terrific effort in coding things up quickly and for fixing bugs. Anand Great news! I did a quick check in the docs directory i.e. https://github.com/gluster/glusterfs/tree/master/doc to see if we have any documentation about nfs-ganesha or pNFS and glusterfs integration, but did not find any. I think without howtos around this will hamper the adoption of this feature among users. So if we can get some documentation for this, it will be awesome. Thanks, Lala Documentation for glusterfs-nfs-ganesha integration is already present : https://forge.gluster.org/nfs-ganesha-and-glusterfs-integration http://blog.gluster.org/2014/09/glusterfs-and-nfs-ganesha-integration/ For pNFS, I will send a documentation as soon as possible. Thanks, Jiffin Forwarded Message Subject: Change in ffilz/nfs-ganesha[next]: pNFS code drop enablement and checkpatch warnings fixed Date: Sat, 21 Mar 2015 01:04:30 +0100 From: GerritHub supp...@gerritforge.com Reply-To: ffilz...@mindspring.com To: Anand Subramanian ana...@redhat.com CC: onnfrhvruutnzhnaq.-g...@noclue.notk.org From Frank Filzffilz...@mindspring.com: Frank Filz has submitted this change and it was merged. Change subject: pNFS code drop enablement and checkpatch warnings fixed .. pNFS code drop enablement and checkpatch warnings fixed Change-Id: Ia8c58dd6d6326f692681f76b96f29c630db21a92 Signed-off-by: Anand Subramanianana...@redhat.com --- A src/FSAL/FSAL_GLUSTER/ds.c M src/FSAL/FSAL_GLUSTER/export.c M src/FSAL/FSAL_GLUSTER/gluster_internal.h M src/FSAL/FSAL_GLUSTER/handle.c M src/FSAL/FSAL_GLUSTER/main.c A src/FSAL/FSAL_GLUSTER/mds.c 6 files changed, 993 insertions(+), 0 deletions(-) -- To view, visithttps://review.gerrithub.io/221683 To unsubscribe, visithttps://review.gerrithub.io/settings Gerrit-MessageType: merged Gerrit-Change-Id: Ia8c58dd6d6326f692681f76b96f29c630db21a92 Gerrit-PatchSet: 1 Gerrit-Project: ffilz/nfs-ganesha Gerrit-Branch: next Gerrit-Owner: Anand Subramanianana...@redhat.com Gerrit-Reviewer: Frank Filzffilz...@mindspring.com Gerrit-Reviewer:onnfrhvruutnzhnaq.-g...@noclue.notk.org ___ Gluster-devel mailing list gluster-de...@gluster.org http://www.gluster.org/mailman/listinfo/gluster-devel ___ Gluster-users mailing list Gluster-users@gluster.org http://www.gluster.org/mailman/listinfo/gluster-users
Re: [Gluster-users] Out of memory Gluster 3.6.2
Could you tell us what activities were run in the cluster? cmd_log_history across all the nodes would give a clear picture of it. ~Atin On 03/24/2015 01:03 PM, Félix de Lelelis wrote: Hi, Today, Glusterd daemon has been killed due to excessive memory consumption: [3505254.762715] Out of memory: Kill process 7780 (glusterd) score 581 or sacrifice child [3505254.763451] Killed process 7780 (glusterd) total-vm:3537640kB, anon-rss:1205240kB, file-rss:672kB I have installed gluster 3.6.2 on centos 7. There in any way to avoid this whithout the need to kill the process?Simply need more memory? 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