| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
can build cpgx to support any of them
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
| |
when that node dies; add todo comments
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
| |
note to stop using join/leave lists
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
| |
the synced_nodes and sync_from weren't accounting for nodes that had
been removed. Problems here found by Honza.
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
|
| |
If an error occurs with debug/event output is disabled (-D0 -H0),
then print the buffer of debug/event output to stderr before exit(1).
The buffer is still written to /var/log/cluster/cpgx_debug.txt.
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
| |
of the cpg
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
| |
wait this many seconds between kill/restart
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
|
| |
it can just be convenient
-IA runs the iptables -A command
-ID runs the iptables -D command
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
| |
stop traffic with iptables prior to killing the corosync process;
I think this tends to look more like a real node failure
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
| |
of corosync or aisexec, and optionally by starting exec directly
instead of via cman_tool
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
| |
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
| |
with -d1 option
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
| |
copying various build-related stuff from Nate,
http://fedorapeople.org/gitweb?p=nstraz/public_git/gxpp.git
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
|
| |
. automate whitetank build stuff
. -i <sec> option to specify how many seconds to run
. -t <sec> option to specify timeout waiting for a dispatch
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
| |
deal with lib64 and whitetank differences
fix some warnings from -Wall
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
|
| |
. send syncs to nodes in order of their joining
. tuning some constants
. rewording, "exit" instead of "fail"
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
test something similar to a node failure, as opposed to just a process
failure, by kill -9 corosync, then cman_tool join
May want to use iptables instead of killing corosync to better approximate
a realistic node failure.
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
|
|
| |
The state of synced_nodes need to be snapshotted at the point of a joining
node's confchg.
Also some code munging.
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
. move code around
. save debug log, write it out on error
. configurable sync_max
. don't save sync history (just header) to others while waiting for our sync
. error if no dispatches for 10 sec
. rework confchg leave/join/sync wait/done state checks
. check for case where all others leave without sending us sync
. check all events after time message
. don't send syncs if we're leaving or left
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
|
|
| |
Signed-off-by: David Teigland <teigland@redhat.com>
|
|
Signed-off-by: David Teigland <teigland@redhat.com>
|