On 07/15/2015 07:39 PM, Gregor Burck wrote:> Hi Ravi,Hi Gregor, Please retain the CC to gluster-users when you reply.> >> That should not be the case, can you provide the client (mount) logs >> and the brick logs when this happens? Replicate translator returns >> EROFS usually when quorum is not met. > Which is the mount log? > Is it this: /var/log/glusterfs/import-vbstore.log ? >Right.> The brick log I think is this: > /var/log/glusterfs/bricks/export-vbstore.logRight again. Please include these logs from all the nodes when you reboot the 3rd node and the VM goes read only.> > Here I see something special, the time is out of sync. > date: Mi 15. Jul 16:04:38 CEST 2015 > log file entry: [2015-07-15 14:03:56.009191] > > The system time should be europe/berlin, but in the log I got > greenwich time? > > When the logfiles right, I clear them, after that I restart an node. > > Thank you for help! > > Gregor >
Hi Ravi, now I could send the logs. What I do: 1. 've a look that heal is clean 2. clear logfiles 3. reboot edgf006 If edgf006 is back I make some filesystem operations in virtual machine. Imediatly I got an error. I wonder, the import-logs still clear? I do an other test with an dedicated client. but here the vm crash, too. Bye Gregor -------------- next part -------------- A non-text attachment was scrubbed... Name: edgf004_export-vbstore.log Type: text/x-log Size: 75415 bytes Desc: not available URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150716/fe5f2399/attachment.bin> -------------- next part -------------- A non-text attachment was scrubbed... Name: edgf005_export-vbstore.log Type: text/x-log Size: 32544 bytes Desc: not available URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150716/fe5f2399/attachment-0001.bin> -------------- next part -------------- A non-text attachment was scrubbed... Name: edgf006_export-vbstore.log Type: text/x-log Size: 76732 bytes Desc: not available URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150716/fe5f2399/attachment-0002.bin>
One Point is, the difference between the log file time (UTC) and system time (europe/Berlin) Could that be an hint? Bye Gregor
Hi, I test now following: 1. start the vm 2. shutdown one node (edgf006) 3. remove-brick edgf006 4. start edgf006 5. add-brick edgf006 6. gluster volume heal vbstore full The VM still running,... That is not a solution but maybee an workaround. Would bee nice if we found a solutions for shuting down a brick and restart it without remove it from cluster. What should I do for an maintenance shutdown of one brick, even for change a fan or other stuff? Is there something like 'gluster volume maintenance-brick <VOLUME> planed? Bye Gregor