URL: <http://savannah.nongnu.org/bugs/?20187> Summary: cluster/unify not giving consistent outputs Project: Gluster Submitted by: ynot Submitted on: Saturday 06/16/2007 at 17:14 Category: GlusterFS Severity: 3 - Normal Priority: 5 - Normal Item Group: Improper behaviour Status: None Privacy: Private Assigned to: None Open/Closed: Open Discussion Lock: Any Operating System: GNU/Linux _______________________________________________________ Details: For this test, I've setup a rr scheduler. It looks like from the FAQ that the alu scheduler has more features and testing done to it. I've created a 250GB data set on two servers. No replication yet. I'm using TCP/IP transport. I create a few directories. I reboot one server. I removed one directory. I reconnect the server. I run ls over and over and over again, and notice that the outputs are different. The general pattern is about every 4th ls I run, I get a different output. I should get the exact output each time. The scenario is that a server has gone down for a while, and then later through the help of IT folks, a new box may or may not be built up. Any *deletions* (Unify seems to handle additions OK, but it is unclear how deletions or moves of files or directories are handled) should be properly handled. The same box or a replacement box may come up. in the case of the same box (maybe no replicas exist), the erasures states should be properly propagated to the old box. In case there is replica data, this case should be properly handled as well. OS: Fedora 7 _______________________________________________________ Reply to this item at: <http://savannah.nongnu.org/bugs/?20187> _______________________________________________ Message sent via/by Savannah http://savannah.nongnu.org/
