similar to: How can replicated and distributed replicated volume benefit read performance in most environments?

Hi Ben, Yes, I was referring to the same link before and I think most of the FOPS I've understood from the definitions there and I'll update if I found any corrections or new definitions. Btw, my previous underatanding about LOOKUP is wrong, its NOT dns lookup. Regards, Jeevan. On Nov 14, 2017 11:02 AM, "Ben Turner" <bturner at redhat.com> wrote: Here is some of the

Hi, Also I forgot to mention there are DUMP and PORTBYBRICK calls too. I understood these too calls are to get updated on gluster configuration. But all these two calls are to a particular node only while there are total (2+1)*3= 6 nodes in my configuration. Why so? What exactly is DUMP and PORTBYBRICK? Regards Jeevan. On Dec 10, 2017 6:47 PM, "Jeevan Patnaik" <g1patnaik at

Filtering the brick logs in TRACE mode with rpcsvc.c does show the FOPS. >From this, I've realized that LOOKUP is actually dns lookup. This actually differs from NFS lookup operation. Please correct me if I'm wrong. Regards, Jeevan. On Nov 13, 2017 9:40 PM, "Jeevan Patnaik" <g1patnaik at gmail.com> wrote: > Hi, > > Can I get a brief description of all the

An ESTALE error usually means the gfid could not be found. Does repeating the "rm -rf" delete the directory? Regards, Nithya On 3 January 2018 at 12:16, Jeevan Patnaik <g1patnaik at gmail.com> wrote: > Hi all, > > I haven't found any root cause or workaround for this yet. Can any one > help me in underatanding the issue? > > Regards, > Jeevan. > >

Hi all, I haven't found any root cause or workaround for this yet. Can any one help me in underatanding the issue? Regards, Jeevan. On Dec 21, 2017 8:20 PM, "Jeevan Patnaik" <g1patnaik at gmail.com> wrote: > Hi, > > > After running rm -rf on a directory, the files under it got deleted, but > the directory was not deleted and was showing stale file handle

Hi, There's no client activity and I have collected tcpdump for 5 mins on one of the glusterFs sever which is also a client with gluster NFS mount. From tcpdump I see that for every few seconds, STAT & STATFS commands are run. For whichctory are these commands being run and why? Regards, Jeevan. -------------- next part -------------- An HTML attachment was scrubbed... URL:

Hi, Can I get a brief description of all the FOPS in gluster or the location of the source code file so that I will try to get an understanding myself? Few FOPS I'm not clear like FORGET, UNLINK, FLUSH, LOOKUP Or is there a way I can tunnel through the FOPS that that are happening in the background for each operation? I have tried this to find from a brick logfile in TRACE mode, but there

trusted.io-stats-dump is a virtual (not physical) extended attribute. The code is written in a way that a request to set trusted.io-stats-dump gets bypassed at the io-stats translator layer on the stack and there it gets converted into the action of dumping the statistics into the provided output file path. See io_stats_setxattr() implementation in io-stats.c for more details. HTH, Krutika On

I'm hitting the 'cannot find stripe size' bug: [2013-04-29 17:42:24.508332] E [stripe-helpers.c:268:stripe_ctx_handle] 0-gv0-stripe-0: Failed to get stripe-size [2013-04-29 17:42:24.513013] W [fuse-bridge.c:968:fuse_err_cbk] 0-glusterfs-fuse: 867: FSYNC() ERR => -1 (Invalid argument) Is there a fix for this in 3.3.1 or do we need to move to git HEAD to make this work? M. --

Hi, I am trying to understand how the extended attribute trusted.io-stats-dump works. setfattr -n trusted.io-stats-dump -v /tmp/gluster_perf_stats/io-stats-pre.txt /mnt/gluster/gv0_glusterfs I can see that the io-stats-pre.txt is created. But how and what happened in the background? And why I can't I see the attribute with getfattr again? getfattr -dm- /mnt/gluster/gv0_glusterfs # file:

Hi, I'm trying to create striped replicated volume but getting tis error: gluster volume create cloud stripe 4 replica 2 transport tcp nebula1:/dataPool nebula2:/dataPool nebula3:/dataPool nebula4:/dataPool wrong brick type: replica, use<HOSTNAME>:<export-dir-abs-path> Usage: volume create<NEW-VOLNAME> [stripe<COUNT>] [replica<COUNT>]

Hi, After running rm -rf on a directory, the files under it got deleted, but the directory was not deleted and was showing stale file handle error. After 18 minutes, I'm able to delete the directory. So could anyone help me in knowing what could have happened or when in general I get such errors. The following is NFS log: [2017-12-21 13:56:01.592256] I [MSGID: 108019]

Hy, I have the following question: If I build a replicated-striped volume (one replica and one stripe) when I want to grow that volume I can grow it adding one brick and its replica or I have to add the stripe and Its replica also? Hope you can help me, thanks in advance Juan Brenes Imprima este correo solo si es necesario. Act?e responsablemente con el Medio Ambiente.

Hi, >From strace, I can see the system calls setxattr and getxattr. Two things I don't understand is: 1. how does the file got created so quickly, from where it read the stats. 2.why is the added attribute missing if I check with getfattr? Regards, Jeevan. On Nov 13, 2017 12:40 PM, "Amar Tumballi" <atumball at redhat.com> wrote: On Mon, Nov 13, 2017 at 12:14 PM,

Hi, We have around 60 hosts and each of them acts as glusterFs clients as well as server. To achieve HA, my underatanding is that we can use Ganesha NFS alone (and not Kernel NFS) and for above 3.10 versions, the HA packages are not ready yet. So, I'm thinking if I can avoid HA by making each client acts as server for itself. i.e.,the server name specified while mounting is localhost. So,

Good afternoon, I am playing around with GlusterFS 3.1 in CentOS 6 virtual machines to see if I can get of proof of concept for a bigger project. In my setup, I have 4 GlusterFS servers with two bricks each of 10GB with XFS (per your quick-start guide). So, I have a total of 8 bricks. When bu I have no problem with distributed-replicated volumes. However, when I set up a striped replicated

On Mon, Nov 13, 2017 at 12:14 PM, Jeevan Patnaik <g1patnaik at gmail.com> wrote: > Hi, > > I am trying to understand how the extended attribute trusted.io-stats-dump > works. > > setfattr -n trusted.io-stats-dump -v /tmp/gluster_perf_stats/io-stats-pre.txt > /mnt/gluster/gv0_glusterfs > > I can see that the io-stats-pre.txt is created. But how and what happened

Hi there, each time I copy (or dd or similar) a file to a striped replicated volume I get an error: the argument is not valid. An empty file is created. If I now run the copy, it works. This is in independed of the client platform. We are using version 3.3.1 Mit freundlichen Gr??en / Kind regards Axel Weber -------------- next part -------------- An HTML attachment was scrubbed... URL:

Hi, What I meant was that every host acts as nfs server for itself alone. A is mounted on A, B is mounted on B etc. So, if a host is crashed or storage of the host is not available, other hosts won't be affected. And how bad can be the performance overhead in having multiple nfs servers? Regards, Jeevan. On Nov 15, 2017 10:07 AM, "Soumya Koduri" <skoduri at redhat.com>

We are having trouble creating a stripe 2 replica 2 volume across 4 hosts: user at gluster-fs-host-0:/gfsr$ sudo gluster volume create sr stripe 2 replica 2 glusterfs-host-0:/gfsr glusterfs-host-1:/gfsr glusterfs-host-2:/gfsr glusterfs-host-3:/gfsr wrong brick type: replica, use <HOSTNAME>:<export-dir-abs-path> We are on glusterfs 3.2.5