Andrew Tridgell [Tue, 5 Jun 2007 07:57:07 +0000 (17:57 +1000)]
clean out some more cruft
Andrew Tridgell [Tue, 5 Jun 2007 07:43:19 +0000 (17:43 +1000)]
first step in health monitoring of cluster nodes. When not healthy they will be marked disabled
Andrew Tridgell [Tue, 5 Jun 2007 05:18:37 +0000 (15:18 +1000)]
set close on exec on pipe in event scripts, so long running scripts don't hold the pipe
Andrew Tridgell [Tue, 5 Jun 2007 05:17:53 +0000 (15:17 +1000)]
more unused code
Andrew Tridgell [Tue, 5 Jun 2007 05:17:24 +0000 (15:17 +1000)]
remove an unused function
Andrew Tridgell [Mon, 4 Jun 2007 13:54:56 +0000 (23:54 +1000)]
use the right IP from the passed structure in takeip/releaseip calls
Andrew Tridgell [Mon, 4 Jun 2007 13:54:46 +0000 (23:54 +1000)]
explain event types
Andrew Tridgell [Mon, 4 Jun 2007 13:54:22 +0000 (23:54 +1000)]
ignore commented out entries in /etc/exports
Andrew Tridgell [Mon, 4 Jun 2007 13:52:12 +0000 (23:52 +1000)]
use the right IP from the passed structure in takeip/releaseip calls
Andrew Tridgell [Mon, 4 Jun 2007 12:28:52 +0000 (22:28 +1000)]
don't crash doing ctdb ip when not doing takeover
Andrew Tridgell [Mon, 4 Jun 2007 12:13:59 +0000 (22:13 +1000)]
merge from ronnie
Ronnie Sahlberg [Mon, 4 Jun 2007 11:11:51 +0000 (21:11 +1000)]
add a control that lists all public ip addresses and which node that
currently serves it
Andrew Tridgell [Mon, 4 Jun 2007 10:22:44 +0000 (20:22 +1000)]
make recovery daemon values tunable
Ronnie Sahlberg [Mon, 4 Jun 2007 10:10:53 +0000 (20:10 +1000)]
merge from tridge
Ronnie Sahlberg [Mon, 4 Jun 2007 10:07:37 +0000 (20:07 +1000)]
change the takoverip/releaseip controls to pass a structure containing
both the nodenumber and the id of the node that has taken over that
address in addition to the public address itself so that all nodes
can learn which node is currently hosting each of the public addresses
Andrew Tridgell [Mon, 4 Jun 2007 10:05:31 +0000 (20:05 +1000)]
allow setting of variables at startup in config file
Andrew Tridgell [Mon, 4 Jun 2007 09:53:19 +0000 (19:53 +1000)]
added tunables settable using ctdb command line tool
Andrew Tridgell [Mon, 4 Jun 2007 07:46:37 +0000 (17:46 +1000)]
- start moving tunable variables into their own structure
- fixed the test scripts to use a separate dbdir
Andrew Tridgell [Mon, 4 Jun 2007 07:23:55 +0000 (17:23 +1000)]
remove some cruft thats not needed any more
Andrew Tridgell [Mon, 4 Jun 2007 05:44:52 +0000 (15:44 +1000)]
make sure we don't have any namespace collision problems with config variables
Ronnie Sahlberg [Mon, 4 Jun 2007 05:44:13 +0000 (15:44 +1000)]
merge from tridge
Andrew Tridgell [Mon, 4 Jun 2007 05:09:03 +0000 (15:09 +1000)]
split out the basic interface handling, and run event scripts in a deterministic order
Andrew Tridgell [Mon, 4 Jun 2007 04:16:51 +0000 (14:16 +1000)]
automatically bring up interfaces that we manage. This allows ctdb to work without requiring two IPs per public interface
Andrew Tridgell [Mon, 4 Jun 2007 03:34:27 +0000 (13:34 +1000)]
merged from ronnie
Ronnie Sahlberg [Mon, 4 Jun 2007 03:31:58 +0000 (13:31 +1000)]
show the second column in the machinereadable output for ctdb status as
IP
Ronnie Sahlberg [Mon, 4 Jun 2007 03:26:07 +0000 (13:26 +1000)]
add the ip address to the nodemap structure we pull from a server and
display the physical address of a node when we do a ctdb status
Ronnie Sahlberg [Mon, 4 Jun 2007 01:48:36 +0000 (11:48 +1000)]
merge from tridge
Andrew Tridgell [Sun, 3 Jun 2007 12:11:48 +0000 (22:11 +1000)]
handle NETWORKING var not existing
Andrew Tridgell [Sun, 3 Jun 2007 12:07:07 +0000 (22:07 +1000)]
make the init scripts more portable about location of system config files
Ronnie Sahlberg [Sun, 3 Jun 2007 11:30:45 +0000 (21:30 +1000)]
merge from tridge
Andrew Tridgell [Sun, 3 Jun 2007 10:54:24 +0000 (20:54 +1000)]
merged from ronnie
Ronnie Sahlberg [Sun, 3 Jun 2007 09:50:51 +0000 (19:50 +1000)]
add a -Y option to generate machine readable output.
print 'ctdb status' in machinereadable form as
:VNN:0|1:
Ronnie Sahlberg [Sun, 3 Jun 2007 09:24:52 +0000 (19:24 +1000)]
ubuntu uses a different style of init scripts than redhat and suse
Ronnie Sahlberg [Sun, 3 Jun 2007 08:59:27 +0000 (18:59 +1000)]
print an error message to stdout if we failed to open the logfile for
the daemon
Ronnie Sahlberg [Sun, 3 Jun 2007 08:41:29 +0000 (18:41 +1000)]
merge from tridge
Andrew Tridgell [Sun, 3 Jun 2007 07:53:26 +0000 (17:53 +1000)]
more portability tweaks in the init script
Andrew Tridgell [Sun, 3 Jun 2007 07:39:11 +0000 (17:39 +1000)]
fixed location of init.d directory to work on SLES and RHEL
Ronnie Sahlberg [Sun, 3 Jun 2007 07:07:23 +0000 (17:07 +1000)]
merge from tridge
Andrew Tridgell [Sun, 3 Jun 2007 06:33:08 +0000 (16:33 +1000)]
doc updates
Andrew Tridgell [Sun, 3 Jun 2007 05:56:23 +0000 (15:56 +1000)]
web page tidy ups
Andrew Tridgell [Sun, 3 Jun 2007 04:41:28 +0000 (14:41 +1000)]
merge from ronnie
Andrew Tridgell [Sun, 3 Jun 2007 04:39:27 +0000 (14:39 +1000)]
don't start nfs services unless the relevant directories are available
Andrew Tridgell [Sun, 3 Jun 2007 03:21:07 +0000 (13:21 +1000)]
test commit
Ronnie Sahlberg [Sun, 3 Jun 2007 01:48:13 +0000 (11:48 +1000)]
mention that ctdb offers cross cluster messaging to applications
Ronnie Sahlberg [Sun, 3 Jun 2007 01:40:50 +0000 (11:40 +1000)]
merged with tridge
update to the web page for ctdb
Andrew Tridgell [Sun, 3 Jun 2007 00:29:57 +0000 (10:29 +1000)]
do a full restart in init cron call
Andrew Tridgell [Sun, 3 Jun 2007 00:29:14 +0000 (10:29 +1000)]
fixed a race condition in the handling of the recovery lock
Ronnie Sahlberg [Sat, 2 Jun 2007 23:58:51 +0000 (09:58 +1000)]
when we get a dmaster error, show the database id in the log so we can
track which db it happens for
Ronnie Sahlberg [Sat, 2 Jun 2007 22:32:29 +0000 (08:32 +1000)]
merge from tridge
Andrew Tridgell [Sat, 2 Jun 2007 09:45:06 +0000 (19:45 +1000)]
docs on how to use statd-callout
Andrew Tridgell [Sat, 2 Jun 2007 09:40:07 +0000 (19:40 +1000)]
move config files to config/ directory
Andrew Tridgell [Sat, 2 Jun 2007 09:23:48 +0000 (19:23 +1000)]
removed some old cruft
Andrew Tridgell [Sat, 2 Jun 2007 08:57:21 +0000 (18:57 +1000)]
ctdb_test.c is gone
Andrew Tridgell [Sat, 2 Jun 2007 08:51:05 +0000 (18:51 +1000)]
add an easy way to setup ctdb to start/stop samba
Andrew Tridgell [Sat, 2 Jun 2007 08:08:16 +0000 (18:08 +1000)]
merge from ronnie
Ronnie Sahlberg [Sat, 2 Jun 2007 06:50:58 +0000 (16:50 +1000)]
fix broken link to the CTDB setup page
Ronnie Sahlberg [Sat, 2 Jun 2007 06:44:15 +0000 (16:44 +1000)]
update the evens scripts for nfs and nfslock to honour CTDB_MANAGES_NFS
which is set in /etc/sysconfig/nfs
Ronnie Sahlberg [Sat, 2 Jun 2007 06:33:17 +0000 (16:33 +1000)]
STATD_SHARED_DIRECTORY should be define din the nfs sysconfig file and
not the ctdb sysconfig file since this variable has nothing at all to do
with ctdb
Andrew Tridgell [Sat, 2 Jun 2007 03:31:36 +0000 (13:31 +1000)]
make test now works again
Andrew Tridgell [Sat, 2 Jun 2007 03:16:11 +0000 (13:16 +1000)]
first step towards fixing "make test" with the new daemon system
Andrew Tridgell [Sat, 2 Jun 2007 01:36:42 +0000 (11:36 +1000)]
- make specification of a recovery lock file compulsory
- die if someone other than the recmaster can get the recovery lock
Andrew Tridgell [Sat, 2 Jun 2007 00:03:28 +0000 (10:03 +1000)]
- moved cmdline options that are only relevant to ctdbd into ctdbd.c
- fixed a valgrind error on failing to send a control
- don't mark node dead when already disconnected
- moved node list lock code into common code
Andrew Tridgell [Sat, 2 Jun 2007 00:01:17 +0000 (10:01 +1000)]
another place we need to cope with the strange epoll fork semantics
Andrew Tridgell [Fri, 1 Jun 2007 22:46:49 +0000 (08:46 +1000)]
disable realtime scheduler in event scripts
Andrew Tridgell [Fri, 1 Jun 2007 22:41:19 +0000 (08:41 +1000)]
- make calling of recovered event script async
- shutdown sockets before calling shutdown script
Andrew Tridgell [Fri, 1 Jun 2007 14:11:53 +0000 (00:11 +1000)]
put nfs events in spec and Makefile.in
Andrew Tridgell [Fri, 1 Jun 2007 14:10:22 +0000 (00:10 +1000)]
added nfs event script
Andrew Tridgell [Fri, 1 Jun 2007 13:33:01 +0000 (23:33 +1000)]
don't strictly need netcat
Andrew Tridgell [Fri, 1 Jun 2007 13:25:33 +0000 (23:25 +1000)]
make the packaging much more portable - tested on SLES9 and RHEL4
Andrew Tridgell [Fri, 1 Jun 2007 11:20:05 +0000 (21:20 +1000)]
- make symlink relative in install
- include ctdb functions in samba and nfslock event scripts
Andrew Tridgell [Fri, 1 Jun 2007 11:10:49 +0000 (21:10 +1000)]
tidy up the install somewhat
Andrew Tridgell [Fri, 1 Jun 2007 10:54:26 +0000 (20:54 +1000)]
split out events for each subsystem separately
Andrew Tridgell [Fri, 1 Jun 2007 09:16:58 +0000 (19:16 +1000)]
use a subdirectory for ctdb state files
Andrew Tridgell [Fri, 1 Jun 2007 09:05:41 +0000 (19:05 +1000)]
make the running of the takeover and release event scripts async, to prevent outages due to slow scripts
Andrew Tridgell [Fri, 1 Jun 2007 09:04:51 +0000 (19:04 +1000)]
convert ctdbd.sh tests to use an event script
Andrew Tridgell [Fri, 1 Jun 2007 07:16:19 +0000 (17:16 +1000)]
better download instructions
Andrew Tridgell [Fri, 1 Jun 2007 06:07:13 +0000 (16:07 +1000)]
added package download
Andrew Tridgell [Fri, 1 Jun 2007 05:33:45 +0000 (15:33 +1000)]
merge initial web site from ronnie
Ronnie Sahlberg [Fri, 1 Jun 2007 05:31:23 +0000 (15:31 +1000)]
add a developers section
Ronnie Sahlberg [Fri, 1 Jun 2007 05:23:57 +0000 (15:23 +1000)]
initial webpage
Andrew Tridgell [Fri, 1 Jun 2007 05:23:16 +0000 (15:23 +1000)]
log dates/time in event startup messages
Andrew Tridgell [Fri, 1 Jun 2007 03:50:18 +0000 (13:50 +1000)]
added CTDB_WAIT_DIRECTORIES support
Andrew Tridgell [Fri, 1 Jun 2007 03:29:11 +0000 (13:29 +1000)]
merged from ronnie
Ronnie Sahlberg [Fri, 1 Jun 2007 03:26:14 +0000 (13:26 +1000)]
it is -f not -x to check if a file exists
Ronnie Sahlberg [Fri, 1 Jun 2007 03:14:05 +0000 (13:14 +1000)]
- create /etc/ctdb/taken_ips and /etc/ctdb/changed_ips analog to the
existing /etc/ctdb/released_ips
- only call the statd-callout script if the ips have changed and call
it with a "notify" argument. we need to restart nfslock service in
both cases
- change statd-callout to explicitely restart the lock manager and statd
when "notify" is called. copy the state directory for each held ip
from shared storage to /tmp then use sm-notify to send notifications to
all monitored clients
Ronnie Sahlberg [Fri, 1 Jun 2007 03:03:53 +0000 (13:03 +1000)]
new branch from tridges tree
Andrew Tridgell [Thu, 31 May 2007 03:50:53 +0000 (13:50 +1000)]
ctdb is GPL not LGPL
Andrew Tridgell [Thu, 31 May 2007 01:22:43 +0000 (11:22 +1000)]
if there is no node available to take an IP, don't consider that an error
Andrew Tridgell [Thu, 31 May 2007 01:14:07 +0000 (11:14 +1000)]
better location for statd-callout
Andrew Tridgell [Thu, 31 May 2007 01:09:45 +0000 (11:09 +1000)]
added hooks to make nfs statd behave correctly on failover
Andrew Tridgell [Wed, 30 May 2007 08:06:57 +0000 (18:06 +1000)]
merge lib/replace from samba4
Andrew Tridgell [Wed, 30 May 2007 06:37:39 +0000 (16:37 +1000)]
tell newly connected nodes about any tcp tickle records that we have that they don't have
Andrew Tridgell [Wed, 30 May 2007 06:11:39 +0000 (16:11 +1000)]
use our own netmask when deciding if we should takeover a IP, not the other nodes
- check if ctdb dies while waiting for the startup event
Andrew Tridgell [Wed, 30 May 2007 05:43:25 +0000 (15:43 +1000)]
close sockets when we exec scripts
Andrew Tridgell [Wed, 30 May 2007 04:46:14 +0000 (14:46 +1000)]
we need to listen at transport initialise stage to find our own node number
Andrew Tridgell [Wed, 30 May 2007 04:35:22 +0000 (14:35 +1000)]
- use a CTDB_BROADCAST_ALL for the attach message so it goes to currently disconnected nodes
- start node monitoring only after transport starts
- check if a node is already disconnected in the node dead function
Andrew Tridgell [Wed, 30 May 2007 03:26:50 +0000 (13:26 +1000)]
don't start the transport connecting to the other nodes until after the startup event script has run
Andrew Tridgell [Wed, 30 May 2007 02:37:03 +0000 (12:37 +1000)]
- nice messages while waiting for tcp services to come up
- added more comments to sysconfig file
Andrew Tridgell [Wed, 30 May 2007 02:27:58 +0000 (12:27 +1000)]
wait for local tcp services like smbd to come up before allowing ctdb to start talking to other nodes
Andrew Tridgell [Wed, 30 May 2007 01:17:52 +0000 (11:17 +1000)]
fixed onnode symlink install