Some additional info:
I read somewhere that there has been a problem when the directory
structures are too deep.
On my file servers there is a depth of 9-10 directories, is this
still a problem?
The error I find in the error log is:
performing entry selfheal on 70764b1d-a022-4489-b7e1-df6073765d66
remote operation failed. Path: (null)
(00000000-0000-0000-0000-000000000000) [No data available]
If I look in .glusterfs at the gfid on both machines I see that some
files are missing on the second server.
But the content of .glusterfs matches what is stored in file
structure, so there is not miss match there.
I have written a script that makes an ls on all files in the
structure on a client.
This triggers a heal and the servers are busy working.
I am not done yet as this takes a while.
But this is not the way you are supposed to heal, right?
Best regarards
Marcus
On 02/05/17 11:12, Marcus wrote:
Hi all!
I have setup a replicated gluster cluster on two identical
machines with replica 2.
I run CentOs 7 and gluster version 3.8.11.
I started out with creating a distributed single node gluster
brick.
When I created the brick there was already about 11TB data in
directory before I created the brick.
We used the distributed brick for a couple of weeks before I had
the other server.
I configured the second server and and created replicated bricks
where the second server was empty on data.
gluster volume add-brick vol replica 2 transport tcp
192.168.1.2:/vol
gluster volume add-brick vol2 replica 2 transport tcp
192.168.1.2:/vol2
Everything went fine and files started to replicate and second
server started to be populated with data.
Now two weeks later they are still out of sync.....second server
is still missing data.
If I run: gluster volume heal vol info
There is still thousands of files that need sync from first
server.
Most of the files are shown as a gfid number and not file names.
I have tried to run: gluster volume heal vol
and: gluster volume heal vol full
It reports back that it starts successfully.
I have rebooted both machines.
Still same status.
If I go to a client and use ls on specific files that are not in
sync they become synchronized.
As in:
ls /mymount/mydir/file1
ls /mymount/mydir/file2
Both file1 and file2 gets synchronized.
But I can not do that on thousands of files, right?
How do I get these bricks synchronized and up and running as exact
replicas?
Many thanks in advance!
Best regards
Marcus
_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://lists.gluster.org/mailman/listinfo/gluster-users
--
Marcus Pedersén
System administrator
Interbull Centre
Department of Animal Breeding & Genetics — SLU
Box 7023, SE-750 07
Uppsala, Sweden
Visiting address:
Room 55614, Ulls väg 26, Ultuna
Uppsala
Sweden
Tel: +46-(0)18-67 1962
|
|
 |
_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://lists.gluster.org/mailman/listinfo/gluster-users
