[SSI-devel] [ ssic-linux-Bugs-1385102 ] fastnode returns a DOWN node
Brought to you by:
brucewalker,
rogertsang
From: SourceForge.net <no...@so...> - 2006-06-02 10:15:51
|
Bugs item #1385102, was opened at 2005-12-19 06:36 Message generated for change (Settings changed) made by rogertsang You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=405834&aid=1385102&group_id=32541 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: Sysadmin Group: v1.9.1 >Status: Closed Resolution: Fixed Priority: 5 Submitted By: Nobody/Anonymous (nobody) Assigned to: Nobody/Anonymous (nobody) Summary: fastnode returns a DOWN node Initial Comment: `fastnode` returns a DOWN node. `fast` attempts to execute on a DOWN node. # cluster -v 1: UP 2: DOWN 3: UP # fastnode 2 # fast echo blah Can't execute 'echo', Node 2 is not available. ---------------------------------------------------------------------- Comment By: Roger Tsang (rogertsang) Date: 2006-06-02 06:15 Message: Logged In: YES user_id=1246761 Applied bug fix http://article.gmane.org/gmane.linux.cluster.ssic.cvs/6745 ---------------------------------------------------------------------- Comment By: Roger Tsang (rogertsang) Date: 2006-01-26 00:29 Message: Logged In: YES user_id=1246761 The reason for this problem is nodedown clean up in the kernel sets DOWN non-initnode's loadlevel "load" value to 0 rather than to 0xfffffff (-1). A joining node's load also gets initialized to 0 at first, but the following changes shouldn't matter. A node's loadlevel "load" during idle is usually greater than 0, if I'm not wrong. diff -u -r1.9 onnode.c --- onnode.c 25 Jan 2006 06:47:06 -0000 1.9 +++ onnode.c 26 Jan 2006 04:44:05 -0000 @@ -413,7 +413,7 @@ new_node = (int)node; new_load = do_loads_get(node); /* Skip non-UP nodes */ - if (new_load == -1) + if (new_load <= 0) continue; DBG(("read new node= %d, lm=%lf\n", new_node, new_load)); ---------------------------------------------------------------------- Comment By: Nobody/Anonymous (nobody) Date: 2005-12-23 04:11 Message: Logged In: NO This bug only affects non-initnodes. # onall fastnode (node 1) 3 (node 3) 2 ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=405834&aid=1385102&group_id=32541 |