I build a glusterfs cluster, and the type is the Striped Replicated Volumes with 4 servers like that:
Volume Name: test-volume
Type: Striped-Replicate
Volume ID: bdb596e6-a7d2-44a4-8791-1b4fdc57469c
Status: Started
Number of Bricks: 1 x 2 x 2 = 4
Transport-type: tcp
Bricks:
Brick1: 10.90.245.18:/exp1
Brick2: 10.90.245.35:/exp2
Brick3: 10.90.245.36:/exp3
Brick4: 10.90.245.37:/exp4
I mount it from a client, while upload a 100G file to test-volume. Each of the file on the server should be 50G. While I uploading the data to cluster I turn down a server of cluster. While the client complete upload file then start the server and start service glusterd.
And then the gluster volume heal info like that:
root@alyrokif:/exp1# gluster volume heal test-volume info
Gathering Heal info on volume test-volume has been successful
Brick 10.90.245.18:/exp1
Number of entries: 0
Brick 10.90.245.35:/exp2
Number of entries: 0
Brick 10.90.245.36:/exp3
Number of entries: 0
Brick 10.90.245.37:/exp4
Number of entries: 0
and storage dir in restarted server like that:
root@xetvmcjg:/exp2# ll -h
total 742M
drwxr-xr-x 3 root root 4.0K Mar 28 16:37 ./
drwxr-xr-x 24 root root 4.0K Mar 27 16:30 ../
drw------- 72 root root 4.0K Mar 28 16:36 .glusterfs/
-rw-r--r-- 2 root root 50G Mar 28 16:37 testfile
There was 50G in other 3 server. But there was only 742MB on the restarted server. So how can I make this 742MB to 50G.
Best Answer
I'm assuming you're using Gluster 3.3
Gluster does a proactive self-heal, though it may need to be prodded to notice this. This can be triggered by accessing the file from a client, which should cause it to (within 10 minutes) notice that a sync is incomplete and start healing.
You can also force a heal:
And to force it to heal everything on the volume regardless of the need to do so (it can take an extremely long time if you have a lot of files)
You've already found the command to track healing progress, though this is also reflected in the gluster log-files.