I all,<br>I have some problem with automatic file translator ( afr). I have two server A e B. Both servers have afr client configured.<br> <br>If I stop one server, for example B, The file system managed from AFR is not available for 30 sec in the server A.<br>
<br> Below there is the gluster-client.vol of one of the server.<br><br> volume TSU-1.localdomain-disk<br> type protocol/client<br> option transport-type tcp/client<br> option remote-host 127.0.0.1<br>
option remote-subvolume disk<br> end-volume<br><br> volume TSU-2.localdomain-disk<br> type protocol/client<br> option transport-type tcp/client<br> option remote-host 10.1.48.51<br> option remote-subvolume disk<br>
option transport-timeout 5<br> end-volume<br><br> volume disk-afr<br> type cluster/afr<br> subvolumes TSU-1.localdomain-disk TSU-2.localdomain-disk<br> option favorite-child TSU-1.localdomain-disk<br>
end-volume<br><br> volume writeback-disk<br> type performance/write-behind<br> option aggregate-size 131072<br> subvolumes disk-afr<br> end-volume<br><br> volume readahead-disk<br> type performance/read-ahead<br>
option page-size 65536<br> option page-count 16<br> subvolumes writeback-disk<br> end-volume<br><br>the server has the following configuration:<br>##### Volume #####<br><br> volume local-disk<br>
type storage/posix<br> option directory /glusterfs/shared<br> end-volume<br><br> volume disk<br> type features/posix-locks<br> subvolumes local-disk<br> end-volume<br><br>##### Access Control #####<br>
<br> volume server<br> type protocol/server<br> option transport-type tcp/server<br> subvolumes disk<br> option auth.ip.disk.allow *<br> end-volume<br><br>Is there a way to avoid this behaviour?<br>
Thanks<br>Maurizio<br>