Daniel van Ham Colchete a écrit :
Vicent,
Although it can seen naive, I can bet this problem is happening at your
brick's posix. From what I understand, GlusterFS has no control over
writing or not writing atime. Can you do a "mount|grep noatime" at your
servers?
If you do this test directly at the storage/posix brick, what happens?
You're right, my fault, I had noatime for the underlying filesystem. I
changed that on all server nodes remounted the partitions and restarted
all gluster servers and clients. Looks fine now :)).
--
Vincent Régnard
vregnard@xxxxxxxxxxxxxxxx
TBS-internet.com
027 630 5902
