[Gluster-users] Missing files with unify 2.0.0rc1
Keith Freedman
freedman at FreeFormIT.com
Fri Jan 16 08:57:47 UTC 2009
I *think* (and I'm sure Anand will correct me if I'm wrong), in the
case of a single brick, gluster doesn't care much about it's extended
attributes, but once another brick/node is involved (in unify or
replicate), then the extended attributes become important.
pre-existing data is generally without Xattr's and so this causes confusion.
My guess is, adding the extended attributes (which happens when you
copy the file into the gluster mountpoint) would solve the problem.
I think there's a wiki that suggests how to add xattrs to the files.
At 12:36 AM 1/16/2009, Filipe Maia wrote:
>Pre-existing.
>The problem does not occur if I remove the unify (if I export just one
>volume and import that one volume).
>
>On Thu, Jan 15, 2009 at 22:38, Anand Avati <avati at zresearch.com> wrote:
> > Filipe,
> > how did you populate data into your volumes? was it pre-exising data
> > or did you copy in all data into a freshly created empty volume?
> >
> > avati
> >
> > 2009/1/16 Filipe Maia <filipe at xray.bmc.uu.se>:
> >> Hi,
> >>
> >> I'm trying to use unify to replace my NFS servers but i have
> some problems.
> >> In my tests I also see about a quarter of the fiels that I see on NFS.
> >> I also get the following errors on my glusterfsd.log:
> >>
> >> Version : glusterfs 2.0.0rc1 built on Jan 15 2009 00:02:28
> >> TLA Revision : glusterfs--mainline--3.0--patch-844
> >> Starting Time: 2009-01-15 13:50:01
> >> Command line : glusterfsd
> >> given volfile
> >> +-----
> >> 1: volume disk
> >> 2: type storage/posix
> >> 3: option directory /homes/davinci
> >> 4: end-volume
> >> 5:
> >> 6: volume disk-rs
> >> 7: type features/filter
> >> 8: option root-squashing enable
> >> 9: subvolumes disk
> >> 10: end-volume
> >> 11:
> >> 12: volume iot
> >> 13: type performance/io-threads
> >> 14: subvolumes disk-rs
> >> 15: option thread-count 4
> >> 16: end-volume
> >> 17:
> >> 18: volume brick
> >> 19: type performance/write-behind
> >> 20: subvolumes iot
> >> 21: option window-size 2MB
> >> 22: option aggregate-size 1MB
> >> 23: end-volume
> >> 24:
> >> 25: # Volume name is server
> >> 26: volume server
> >> 27: type protocol/server
> >> 28: option transport-type tcp
> >> 29: option auth.addr.brick.allow *
> >> 30: subvolumes brick
> >> 31: end-volume
> >> +-----
> >>
> >> 2009-01-15 13:50:01 W [xlator.c:382:validate_xlator_volume_options]
> >> brick: option 'aggregate-size' i
> >> s deprecated, preferred is 'block-size', continuing with correction
> >> 2009-01-15 13:50:01 W [xlator.c:382:validate_xlator_volume_options]
> >> brick: option 'window-size' is d
> >> eprecated, preferred is 'cache-size', continuing with correction
> >> 2009-01-15 14:01:42 E [socket.c:104:__socket_rwv] server: readv failed
> >> (Connection reset by peer)
> >> 2009-01-15 14:01:42 E [socket.c:566:socket_proto_state_machine]
> >> server: socket read failed (Connecti
> >> on reset by peer) in state 1 (192.168.1.235:1020)
> >> 2009-01-15 14:04:07 W [posix.c:1042:posix_link] disk: link
> >> /filipe/.Xauthority-n to /filipe/.Xauthor
> >> ity failed: File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/.dbus: File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/.dbus/session-bus: File exist
> >> s
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep: File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep/Defaults: File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep/Library: File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep/Library/WindowMaker:
> >> File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep/Library/WindowMaker/B
> >> ackgrounds: File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep/Library/WindowMaker/I
> >> conSets: File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep/Library/WindowMaker/S
> >> oundSets: File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep/Library/WindowMaker/P
> >> ixmaps: File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep/Library/Icons: File e
> >> xists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep/Library/WindowMaker/S
> >> ounds: File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep/Library/WindowMaker/S
> >> tyles: File exists
> >> 2009-01-15 14:04:08 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/GNUstep/Library/WindowMaker/T
> >> hemes: File exists
> >> 2009-01-15 14:08:57 E [socket.c:104:__socket_rwv] server: readv failed
> >> (Connection reset by peer)
> >> 2009-01-15 14:08:57 E [socket.c:566:socket_proto_state_machine]
> >> server: socket read failed (Connecti
> >> on reset by peer) in state 1 (192.168.1.235:1019)
> >> 2009-01-15 15:34:44 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/.kde: File exists
> >> 2009-01-15 15:34:46 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/.kde/share: File exists
> >> 2009-01-15 15:34:46 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/.kde/share/config: File exist
> >> s
> >> 2009-01-15 15:34:46 W [posix.c:928:posix_symlink] disk: symlink of
> >> /filipe/.kde/socket-gauguin --> /
> >> tmp/ksocket-filipe: File exists
> >> 2009-01-15 15:34:46 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/.qt: File exists
> >> 2009-01-15 15:34:47 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/.kde/share/apps: File exists
> >> 2009-01-15 15:35:02 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/.mcop: File exists
> >> 2009-01-15 15:35:57 W [posix.c:796:posix_mkdir] disk: mkdir of
> >> /filipe/.ssh: File exists
> >> 2009-01-15 15:37:58 E [socket.c:104:__socket_rwv] server: readv failed
> >> (Connection reset by peer)
> >> 2009-01-15 15:37:58 E [socket.c:566:socket_proto_state_machine]
> >> server: socket read failed (Connecti
> >> on reset by peer) in state 1 (192.168.1.235:1016)
> >> 2009-01-15 16:04:32 E [socket.c:104:__socket_rwv] server: writev
> >> failed (Connection reset by peer)
> >> 2009-01-15 16:05:16 E [write-behind.c:1150:wb_flush] brick:
> returning EBADFD
> >> 2009-01-15 16:05:17 E [write-behind.c:1150:wb_flush] brick:
> returning EBADFD
> >> 2009-01-15 16:05:17 E [write-behind.c:1150:wb_flush] brick:
> returning EBADFD
> >> 2009-01-15 16:05:17 E [write-behind.c:1150:wb_flush] brick:
> returning EBADFD
> >> 2009-01-15 16:05:17 E [write-behind.c:1150:wb_flush] brick:
> returning EBADFD
> >> 2009-01-15 16:05:17 E [write-behind.c:1150:wb_flush] brick:
> returning EBADFD
> >> 2009-01-15 16:05:17 E [write-behind.c:1150:wb_flush] brick:
> returning EBADFD
> >> 2009-01-15 16:05:17 E [write-behind.c:1150:wb_flush] brick:
> returning EBADFD
> >>
> >>
> >> I don't think I have any hardware problems as I can cat all the files
> >> in my home directory without any problem.
> >>
> >> I tried to reproduce the problem with a smaller setup without much
> >> luck unfortunately.
> >>
> >> Here is the client file:
> >>
> >> volume tintoretto
> >> type protocol/client
> >> option transport-type tcp
> >> option remote-host tintoretto
> >> option remote-subvolume brick
> >> end-volume
> >>
> >> volume giotto
> >> type protocol/client
> >> option transport-type tcp
> >> option remote-host giotto
> >> option remote-subvolume brick
> >> end-volume
> >>
> >> volume michelangelo
> >> type protocol/client
> >> option transport-type tcp
> >> option remote-host michelangelo
> >> option remote-subvolume brick
> >>
> >> volume donatello
> >> type protocol/client
> >> option transport-type tcp
> >> option remote-host donatello
> >> option remote-subvolume brick
> >> end-volume
> >>
> >> volume ns
> >> type protocol/client
> >> option transport-type tcp
> >> option remote-host tintoretto
> >> option remote-subvolume ns
> >> end-volume
> >>
> >> volume bricks
> >> type cluster/unify
> >> option namespace ns # this will not be storage child of unify.
> >> subvolumes tintoretto michelangelo giotto donatello
> >> # option self-heal foreground # foreground off # default is foreground
> >> # option self-heal background # foreground off # default is foreground
> >> ### ** Round Robin (RR) Scheduler **
> >> option scheduler rr
> >> # A server is not used if it's free disk space drops below 15%.
> >> option scheduler.limits.min-free-disk 15% #%
> >> end-volume
> >>
> >> volume bricks-rs
> >> type features/filter
> >> option root-squashing enable
> >> subvolumes bricks
> >> end-volume
> >>
> >> volume iot
> >> type performance/io-threads
> >> subvolumes bricks-rs
> >> option thread-count 4
> >> end-volume
> >>
> >> volume wb
> >> type performance/write-behind
> >> subvolumes iot
> >> option flush-behind off # default value is 'off'
> >> option window-size 2MB
> >> option aggregate-size 1MB # default value is 0
> >> end-volume
> >>
> >> ### 'IO-Cache' translator is best used on client side when a
> filesystem has file
> >> # which are not modified frequently but read several times. For
> example, while
> >> # compiling a kernel, *.h files are read while compiling every *.c
> >> file, in these
> >> # case, io-cache translator comes very handy, as it keeps the whole
> >> file content in
> >> # the cache, and serves from the cache.
> >> # One can provide the priority of the cache too.
> >>
> >> volume ioc
> >> type performance/io-cache
> >> subvolumes wb
> >> option page-size 1MB # 128KB is default
> >> option cache-size 64MB # 32MB is default
> >> option cache-timeout 5 # 1second is default
> >> option priority *.c:2,*.h:1 # default is *:0
> >> end-volume
> >>
> >>
> >> ### 'Read-Ahead' translator is best utilized on client side, as
> it prefetches
> >> # the file contents when the first read() call is issued.
> >> volume ra
> >> type performance/read-ahead
> >> subvolumes ioc
> >> option page-size 1MB # default is 256KB
> >> option page-count 4 # default is 2
> >> option force-atime-update no # defalut is 'no'
> >> end-volume
> >>
> >>
> >> Filipe
> >>
> >> _______________________________________________
> >> Gluster-users mailing list
> >> Gluster-users at gluster.org
> >> http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users
> >>
> >
>
>_______________________________________________
>Gluster-users mailing list
>Gluster-users at gluster.org
>http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users
More information about the Gluster-users
mailing list