sahlberg/ctdb.git
13 years agoWe default to non-deterministic ip now where ips are "sticky" and dont change 1.2-nodeflags
Ronnie Sahlberg [Wed, 2 Feb 2011 04:00:53 +0000 (15:00 +1100)]
We default to non-deterministic ip now where ips are "sticky" and dont change
too much.
This means we can simplify the way we add ips significantly and stop
trying to move them.

We also check if the node already hosts the ip, in which case we used to return an error. Instead just print an error string but return 0, ok.
This makes it easier to script, and works around broken scripts.

CQ1021034

13 years agoIf the node is stopped, put a log entry in /var/log/* to indicate this is why we...
Ronnie Sahlberg [Mon, 31 Jan 2011 06:40:26 +0000 (17:40 +1100)]
If the node is stopped,   put a log entry in /var/log/* to indicate this is why we never become ready

13 years agoLockWait congestion.
Ronnie Sahlberg [Mon, 24 Jan 2011 00:42:50 +0000 (11:42 +1100)]
LockWait congestion.

Add a dlist to track all active lockwait child processes.
Everytime creating a new lockwait handle, check if there is already an
active lockwait process for this database/key and if so,
send the new request straight to the overflow queue.

This means we will only have one active lockwaic child process for a certain key,
even if there were thousands of fetch-lock requests for this key.

When the lockwait processing finishes for the original request, the processing in d_overflow() will automagically process all remaining keys as well.

Add back a --nosetsched argument to make it easier to run under gdb

13 years agoCompile fix
Ronnie Sahlberg [Sun, 23 Jan 2011 22:43:45 +0000 (09:43 +1100)]
Compile fix

13 years agoctdb_lockwait: create overflow queue.
Rusty Russell [Fri, 21 Jan 2011 10:47:02 +0000 (21:17 +1030)]
ctdb_lockwait: create overflow queue.

Once we have more than 200 children waiting on a particular db, don't create
any more.  Just put them on an overflow queue, and when a child gets a lock
search that queue to see if others were after the same lock (they probably
were).

13 years agoAdd a new test tool that fetch locks a record and then blocks until it receives
Ronnie Sahlberg [Sun, 23 Jan 2011 20:39:33 +0000 (07:39 +1100)]
Add a new test tool that fetch locks a record and then blocks until it receives
user input to unlock the record again.

13 years agoctdb: hold transaction locks during freeze, mark during recover.
Ronnie Sahlberg [Tue, 18 Jan 2011 02:33:24 +0000 (13:33 +1100)]
ctdb: hold transaction locks during freeze, mark during recover.

Make the ctdb parent "mark" the transaction lock once the child process
has frozen/locked the entire database.
This stops the ctdb daemon from using  a blocking fcntl() locking on the tdb during the
read traverse during recovery.

CQ 1021388

13 years agotdb: expose transaction lock infrastructure for ctdb
Rusty Russell [Tue, 18 Jan 2011 00:17:11 +0000 (10:47 +1030)]
tdb: expose transaction lock infrastructure for ctdb

tdb_traverse_read() grabs the transaction lock.  This can cause ctdbd
(which uses it) to block when it should not; expose mark and normal
variants of this lock, so ctdbd's child (the recovery daemon) can
acquire it and the ctdbd parent can mark it was held.

13 years agochange Christinas previous patch to only perform the check/logging
Ronnie Sahlberg [Mon, 17 Jan 2011 01:00:18 +0000 (12:00 +1100)]
change Christinas previous patch to only perform the check/logging
if we are the main ctdb daemon.
Other daemons/child processes are not guaranteed to get events on regular basis
so those should not be checked.

13 years agoimprove timing issue detections
Christian Ambach [Fri, 14 Jan 2011 12:55:28 +0000 (13:55 +0100)]
improve timing issue detections

the original "Time jumped" messages are too coarse to interpret
exactly what was going wrong inside of CTDB.

This patch removes the original logs and adds two other logs that
differentiate between the time it took to work on an event and
the time it took to get the next event.

13 years agoLIBCTDB: add support for traverse
Ronnie Sahlberg [Fri, 14 Jan 2011 06:35:31 +0000 (17:35 +1100)]
LIBCTDB: add support for traverse

13 years agoWe can not always rely on the recovery daemon pinging us in a timely manner
Ronnie Sahlberg [Thu, 13 Jan 2011 22:46:04 +0000 (09:46 +1100)]
We can not always rely on the recovery daemon pinging us in a timely manner
so we need a "ticker" in the main ctdbd daemon too to ensure we get at least one event to process every second.

This will improve the accuracy of "Time jumped" messages and remove false positives when the recovery daemon is "slow".

13 years agoADDIP failure
Ronnie Sahlberg [Thu, 13 Jan 2011 05:17:43 +0000 (16:17 +1100)]
ADDIP failure

Found during automatic regression testing.
We do not allow the takeip/releaseip events to be executed during a recovery.

All of "ctdb addip, ctdb delip, ctdb moveip" use and force these events to
trigger to perform the ip assignments required.

If these commands collide with a recovery, these commands could fail since we do
not allow takeip/releaseip events to trigger during the recovery.
While it is easy to just try running hte command again, this is suboptimal for script use.

Change these commands to retry these operations a few times until either successfull or until we give up.
This makes the commands much easier to use in scripts.

13 years agoIPALLOCATION : If the node is held pinned down in "init" state
Ronnie Sahlberg [Wed, 12 Jan 2011 22:35:37 +0000 (09:35 +1100)]
IPALLOCATION : If the node is held pinned down in "init" state
by external services failing to start, or blocking CTDBD from finishing the startup phase,
we can encounter a situation where we have not yet fully initialized, but a
remote recovery master tries to release a certain ip clusterwide.

In this situation the node that is pinned down in init/startup phase
would fail to perform the release of the ip address since we are not yet fully operational and not yet host any valid interfaces.

In this situation, we just need to remain unhealthy, there is on need to
also ban the node.

Remove the autobanning for this condition and just let the node remain in
unhealthy mode.
Banning is overkill in this situation when the system is broken and just
draws attention to ctdbd instead of the root cause.

13 years agoEventscripts: lower the fail/restart limits for nfsd.
Martin Schwenke [Tue, 11 Jan 2011 06:13:57 +0000 (17:13 +1100)]
Eventscripts: lower the fail/restart limits for nfsd.

We were potentially leaving a node unable to serve requests for too
long.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoEventscripts: use "startstop_nfs restart" to reconfigure NFS.
Martin Schwenke [Tue, 11 Jan 2011 06:13:06 +0000 (17:13 +1100)]
Eventscripts: use "startstop_nfs restart" to reconfigure NFS.

This was defaulting to just "service nfs restart", which doesn't have
the workarounds we need.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoEventscripts: only autostart during a monitor event.
Martin Schwenke [Tue, 11 Jan 2011 06:12:03 +0000 (17:12 +1100)]
Eventscripts: only autostart during a monitor event.

Otherwise we might short-circuit events that are run only once and
actually need to do something.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoEventscripts: print a message when reconfiguring a service.
Martin Schwenke [Tue, 11 Jan 2011 06:10:55 +0000 (17:10 +1100)]
Eventscripts: print a message when reconfiguring a service.

Otherwise there can be strange error messages from services
stopping/starting, without any context.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoEventscripts: work around NFS restart failure under load.
Martin Schwenke [Tue, 11 Jan 2011 06:06:48 +0000 (17:06 +1100)]
Eventscripts: work around NFS restart failure under load.

"service nfs restart" can fail.  To stop nfsd it sends a SIGINT and
nfsd might take a while to process it if the system is loaded.
Starting nfsd may then fail because resources are still in use.

This does some /proc magic to tell nfsd to do no more processing.  It
then runs service stop, kills nfsd with SIGKILL, and then runs service
start.  This is much less likely to fail.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoTYPO
Ronnie Sahlberg [Tue, 11 Jan 2011 05:17:06 +0000 (16:17 +1100)]
TYPO

13 years agoSTATD is 100027 not 1000247
Ronnie Sahlberg [Tue, 11 Jan 2011 05:15:41 +0000 (16:15 +1100)]
STATD is 100027    not 1000247

13 years agoLIBCTDB uninitialized inqueue element
Ronnie Sahlberg [Mon, 10 Jan 2011 20:37:17 +0000 (07:37 +1100)]
LIBCTDB uninitialized inqueue element

From Michael Anderson,
initialize the inqueue element of the ctdb structure to NULL,
else it might be used uninitialized and cause a segv.

13 years agorecoverd: avoid triggering a full recovery if just some ip allocation
Ronnie Sahlberg [Mon, 10 Jan 2011 05:51:56 +0000 (16:51 +1100)]
recoverd: avoid triggering a full recovery if just some ip allocation
has failed.
We dont need to rebuild the databases in this situation, we just
need to try again to sort out the ip address allocations.

13 years agoAdd ctdb_fork(0 which will fork a child process and drop the real-time
Ronnie Sahlberg [Mon, 10 Jan 2011 02:57:49 +0000 (13:57 +1100)]
Add ctdb_fork(0 which will fork a child process and drop the real-time
scheduler for the child.

Use ctdb_fork() from callers where we dont want the child to be running
at real-time privilege.

13 years agoRevert scheduling back to use real-time processes
Ronnie Sahlberg [Mon, 10 Jan 2011 02:35:39 +0000 (13:35 +1100)]
Revert scheduling back to use real-time processes

Revert this patch:
commit 482c302d46e2162d0cf552f8456bc49573ae729d

We may need to use real-time processes for the main daemon and the recovery daemon to handle the cases where systems come under very high loads.

13 years ago60.nfs Check if we have rpc.statd and if not, skip checking for statd
Ronnie Sahlberg [Thu, 6 Jan 2011 04:42:45 +0000 (15:42 +1100)]
60.nfs Check if we have rpc.statd and if not, skip checking for statd
availability at all (since we cant restart it, there is not point checking
if it is alive)

13 years ago41.HTTPD
Ronnie Sahlberg [Tue, 21 Dec 2010 23:27:53 +0000 (10:27 +1100)]
41.HTTPD

Httpd can be very slow to start on some platforms,
wait 5 monitor intervals before we try to restart it if
it has not bound to port 80 yet.
After 10 failed intervals, flag the node as unhealthy.

13 years ago60.nfs
Ronnie Sahlberg [Tue, 21 Dec 2010 23:09:35 +0000 (10:09 +1100)]
60.nfs

Try to restart LOCKD after 10 failures and
flag the node as unhealthy after 15 failures

13 years agoDont run net serverid wipe in the background
Ronnie Sahlberg [Tue, 21 Dec 2010 23:05:40 +0000 (10:05 +1100)]
Dont run net serverid wipe in the background

13 years ago50.samba
Ronnie Sahlberg [Tue, 14 Dec 2010 10:17:14 +0000 (21:17 +1100)]
50.samba

Net serverid wipe can take a bit of time sometimes so background it.

Only perform auto start/stop of the managed service on the monitor event

13 years agoctdb addip:
Ronnie Sahlberg [Mon, 13 Dec 2010 01:06:01 +0000 (12:06 +1100)]
ctdb addip:

After finishing "ctdb addip"  wait for an implicit "iptakeover" to complete
the assignment to a node.

This makes it more wasteful and timeconsuming when adding multiple ips
at once, or the same ip to multiple nodes,
but makes it easier to script the use of this command.

13 years agoLVS
Ronnie Sahlberg [Sun, 12 Dec 2010 08:38:39 +0000 (19:38 +1100)]
LVS

update lvs configuration on ipreallocated events too

13 years agoWhen assigning the single-public-ip during startup,
Ronnie Sahlberg [Sun, 12 Dec 2010 03:22:20 +0000 (14:22 +1100)]
When assigning the single-public-ip during startup,
flag the interface as initially being "link ok"
so that we can add it and startup.

The eventscript can later drop the flag if required

13 years agoRevert "server: when we migrate off a record with data, set the MIGRATED_WITH_DATA...
Ronnie Sahlberg [Mon, 13 Dec 2010 03:22:07 +0000 (14:22 +1100)]
Revert "server: when we migrate off a record with data, set the MIGRATED_WITH_DATA flag"

This reverts commit bb8d975376d959689e702c8e349e787fbde0db01.

13 years agoRevert "Add a new header flag for "migrated with data" and set this to 1"
Ronnie Sahlberg [Mon, 13 Dec 2010 03:21:50 +0000 (14:21 +1100)]
Revert "Add a new header flag for "migrated with data" and set this to 1"

This reverts commit 4a8590f76c5e7ea68acb4d5e52ec0e8a41b65fda.

13 years agolibctdb
Ronnie Sahlberg [Fri, 10 Dec 2010 03:18:28 +0000 (14:18 +1100)]
libctdb

fix a compile problem after renaming a structure field

13 years agoLibCTDB
Ronnie Sahlberg [Fri, 10 Dec 2010 02:39:18 +0000 (13:39 +1100)]
LibCTDB

Add an input queue where we keep received pdus we have not yet processed
This allows us to perform SYNC calls from an ASYNC callback

13 years agoonly run "serverid wipe" if we are actually running samba.
Ronnie Sahlberg [Wed, 8 Dec 2010 00:08:19 +0000 (11:08 +1100)]
only run "serverid wipe" if we are actually running samba.
we dont need to run this on systems where we do run winbind but not samba

13 years agoidtree: fix overflow for v. large ids on allocation and removal
Rusty Russell [Mon, 6 Dec 2010 03:22:38 +0000 (13:52 +1030)]
idtree: fix overflow for v. large ids on allocation and removal

(Imported from SAMBA commit 09a6538969ac).

Chris Cowan tracked down a SEGV in sub_alloc: idp->level can actually
be equal to 7 (MAX_LEVEL) there, as it can be in sub_remove.

(We unfairly blamed a shift of a signed var for this crash in commit
 2db1987f5a3a).

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
13 years agoAdd a new header flag for "migrated with data" and set this to 1
Ronnie Sahlberg [Mon, 6 Dec 2010 05:09:38 +0000 (16:09 +1100)]
Add a new header flag for "migrated with data" and set this to 1
when we migrate a non-empty record onto the node
or a non-empty record off the node

When we migrate a record back to the lmaster and yield the dmaster role,
inspect this flag if if it is still not set, we can delete the record from
the local database as soon as we have migrated it back to the lmaster.

13 years agoadd new command line functions
Ronnie Sahlberg [Mon, 6 Dec 2010 05:07:55 +0000 (16:07 +1100)]
add new command line functions
ctdb readkey <dbid> <key>
ctdb writekey <dbid> <key> <value>

these are mainly intended for debugging of databases and dmaster migration issues

13 years agoadd a new ctdb_ltdb function to delete a record in a normal database
Ronnie Sahlberg [Mon, 6 Dec 2010 05:06:20 +0000 (16:06 +1100)]
add a new ctdb_ltdb function to delete a record in a normal database

13 years agoserver: when we migrate off a record with data, set the MIGRATED_WITH_DATA flag
Michael Adam [Fri, 3 Dec 2010 14:21:51 +0000 (15:21 +0100)]
server: when we migrate off a record with data, set the MIGRATED_WITH_DATA flag

13 years agoAdd 60.ganesha to what gets installed by make install as well as by the RPM
Ronnie Sahlberg [Mon, 6 Dec 2010 00:30:24 +0000 (11:30 +1100)]
Add 60.ganesha to what gets installed by make install as well as by the RPM

13 years agoadd a missing part of the import of the previous ganesha patch
Ronnie Sahlberg [Mon, 6 Dec 2010 00:26:43 +0000 (11:26 +1100)]
add a missing part of the import of the previous ganesha patch

13 years agomake changes to ctdb event scripts to support NFS-Ganesha.
Chandra Seetharaman [Fri, 3 Dec 2010 23:26:22 +0000 (15:26 -0800)]
make changes to ctdb event scripts to support NFS-Ganesha.

make changes to ctdb event scripts to support NFS-Ganesha.

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
13 years agoduring ip allocation, there are failure modes where a node might hold a ip address
Ronnie Sahlberg [Fri, 3 Dec 2010 02:28:35 +0000 (13:28 +1100)]
during ip allocation, there are failure modes where a node might hold a ip address
but thinks it is still unassigned (-1).

add code to the recovery daemon to detect this case and trigger a reallocation
so that the ip gets covered

and change the takeip code to allow for this condition, taking on an ip address that is
already hosted.

cq s1021073

13 years agodont try starting samba through the "init" event
Ronnie Sahlberg [Thu, 2 Dec 2010 19:07:03 +0000 (06:07 +1100)]
dont try starting samba through the "init" event

13 years agoWhen we are no longer the natgw master, dont put the natgw ip on loopback.
Ronnie Sahlberg [Mon, 29 Nov 2010 01:39:14 +0000 (12:39 +1100)]
When we are no longer the natgw master, dont put the natgw ip on loopback.
We put the ip on loopback just to make sure we would still interoperate with
non-standard configurations on unix-KDC, that are configured to verify the optional
HostAddresses field.
This is not required for AD, since AD does not use this field, and is replaced in
unix land with other/better mechanisms than this "dodgy" check.

This makes it "easier" for applications that have bound to the natgw address
to detect a socket problem and try to reconnect/recover if the ip address
is completely missing from the system.

At the same time, use the winbind specific hook that exists to explicitely tell winbindd : this address is gone, so if you have bound to it, this is a good time to close and rebind your socket.

cq 1020333

13 years agoupdate autostart/stop to work for samba
Ronnie Sahlberg [Thu, 18 Nov 2010 04:40:19 +0000 (15:40 +1100)]
update autostart/stop to work for samba

13 years agoadd an explicit _is_managed_service to iscsi eventscript
Ronnie Sahlberg [Thu, 18 Nov 2010 03:15:18 +0000 (14:15 +1100)]
add an explicit _is_managed_service to iscsi eventscript

13 years agoDont pollute the logs with a "file not found" message
Ronnie Sahlberg [Thu, 18 Nov 2010 02:52:46 +0000 (13:52 +1100)]
Dont pollute the logs with a "file not found" message

CQ S1020745

13 years ago60.nfs eventscript should do nothing if NFS isn't managed by CTDB.
Martin Schwenke [Thu, 18 Nov 2010 02:23:40 +0000 (13:23 +1100)]
60.nfs eventscript should do nothing if NFS isn't managed by CTDB.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoEventscript functions - catch failures in ctdb_service_start().
Martin Schwenke [Thu, 18 Nov 2010 00:27:10 +0000 (11:27 +1100)]
Eventscript functions - catch failures in ctdb_service_start().

ctdb_service_start() currently succeeds if ctdb_counter_init()
succeeds.

This changes it to fail when a service start fails.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years ago50.samba eventscript should stop/start services when they become (un)managed.
Martin Schwenke [Thu, 18 Nov 2010 00:04:52 +0000 (11:04 +1100)]
50.samba eventscript should stop/start services when they become (un)managed.

When the value of $CTDB_MANAGES_SAMBA or $CTDB_MANAGES_WINBIND (or
corresponding changes are made to $CTDB_MANAGED_VERSIONS), the
associated service should be started or stopped as necessary.

This add calls to ctdb_start_stop_service() to manage
starting/stopping samba and winbind.

An associated cleanup is made to the initial checks that one of
$CTDB_MANAGES_SAMBA or $CTDB_MANAGES_WINBIND is set, replacing them
with calls to is_ctdb_managed_service().

To handle the winbind cases ctdb_start_stop_service() and
is_ctdb_managed_service() are updated to take an optional service name
parameter.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoadd a new support function ctdb_check_counter_equal()
Ronnie Sahlberg [Wed, 17 Nov 2010 02:50:56 +0000 (13:50 +1100)]
add a new support function ctdb_check_counter_equal()

update nfs to try to restart the service after 10 consecutive failures
and to flag the node unhealthy after 15

add similar function to mountd

13 years agoEventscripts: make loadconfig() function hookable by the test suite.
Martin Schwenke [Tue, 31 Aug 2010 07:40:40 +0000 (17:40 +1000)]
Eventscripts: make loadconfig() function hookable by the test suite.

Rename loadconfig() to _loadconfig().  Add a new loadconfig() that
simply calls _loadconfig().

This makes it easy for the test suite to override loadconfig().

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoMake a time comparison in 60.nfs eventscript more readable.
Martin Schwenke [Tue, 16 Nov 2010 08:42:31 +0000 (19:42 +1100)]
Make a time comparison in 60.nfs eventscript more readable.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years ago60.nfs only fails or warns after 10 consecutive nfsd/statd failures.
Martin Schwenke [Tue, 16 Nov 2010 08:31:18 +0000 (19:31 +1100)]
60.nfs only fails or warns after 10 consecutive nfsd/statd failures.

These failures are sometimes the result of slow restarts so we want to
avoid dirtying the logs or marking a node unhealthy because of them,
unless they are excessive.

For these 2 cases we use the existing fail counting code but hack a
temporary service_name in a subshell to allow separate fail counts.

We also update ctdb_check_rpc() so that it captures the error output
from rpcinfo and we add a message including the service name to the
beginning.  The error is printed to stdout but is also stored in
ctdb_check_rpc_out to allow it to be conditionally used by the caller.
This function also now returns non-zero rather than exiting on
failure.

Other direct rpcinfo calls are relaced by called to ctdb_check_rpc()
for consistency.

Option handling code for service restarts is cleaned up so that fits
in 80 columns.  A more informative restart messageis now used in all
cases, printing the exact command being used to start a service.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoTest suite: fix typo in ctdb ping test grep pattern.
Martin Schwenke [Tue, 12 Oct 2010 00:10:38 +0000 (11:10 +1100)]
Test suite: fix typo in ctdb ping test grep pattern.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoTest suite: match changed output for ctdb ping to disconnected node.
Martin Schwenke [Wed, 6 Oct 2010 05:32:22 +0000 (16:32 +1100)]
Test suite: match changed output for ctdb ping to disconnected node.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoTest suite: make statistics test cope with changes to statistics output.
Martin Schwenke [Fri, 15 Oct 2010 04:09:08 +0000 (15:09 +1100)]
Test suite: make statistics test cope with changes to statistics output.

Signed-off-by: Martin Schwenke <martin@meltin.net>
13 years agoinitialize the statistics to the current time, not start of epoch
Ronnie Sahlberg [Mon, 15 Nov 2010 05:30:44 +0000 (16:30 +1100)]
initialize the statistics to the current time, not start of epoch
this makes "ctdb statistics" show correct "start of starts collection"

13 years agoDont exit the update ip function if the old and new interfaces are the same
Ronnie Sahlberg [Wed, 10 Nov 2010 03:47:28 +0000 (14:47 +1100)]
Dont exit the update ip function if the old and new interfaces are the same
since if they are the same for whatever reason this triggers the system
to go into an infinite loop and is unrobust

The scriptds have been changed instead to be able to cope with this
situation for enhanced robustness

During takeover_run and when merging all ip allocations across the cluster
try to kepe track of when and which node currently hosts an ip address
so that we avoid extra ip failovers between nodes

13 years agochange the takeover script timeout to 9 seconds from 5
Ronnie Sahlberg [Wed, 10 Nov 2010 03:46:45 +0000 (14:46 +1100)]
change the takeover script timeout to 9 seconds from 5

13 years agoDont check remote ip allocation if public ip mgmt is disabled
Ronnie Sahlberg [Wed, 10 Nov 2010 03:46:05 +0000 (14:46 +1100)]
Dont check remote ip allocation if public ip mgmt is disabled

13 years agothis stuff is just so fragile that it will enter infinite recovery and fail loops
Ronnie Sahlberg [Wed, 10 Nov 2010 03:45:43 +0000 (14:45 +1100)]
this stuff is just so fragile  that it will enter infinite recovery and fail loops
on any kind of tiny unexpected error

unconditionally try to remove ip addresses from both old and new interface
before trying to add it to the new interface to make it less
fragile

13 years agodelete from old interface before adding to new interface
Ronnie Sahlberg [Wed, 10 Nov 2010 03:40:43 +0000 (14:40 +1100)]
delete from old interface before adding to new interface
this stops the script from failing with an error if
both interfaces are specified as the same, which otherwise breaks and leads to an infinite recovery loop

13 years agodelay loading the public ip address file until after we have started the transport...
Ronnie Sahlberg [Wed, 10 Nov 2010 01:59:25 +0000 (12:59 +1100)]
delay loading the public ip address file until after we have started the transport and discovered ouw own pnn number

13 years agowhen we load the public address file, at the same time check if we are already hosti...
Ronnie Sahlberg [Wed, 10 Nov 2010 01:11:11 +0000 (12:11 +1100)]
when we load the public address file,  at the same time check if we are already hosting the public address, if so, set ourselves up as the pnn for that address

13 years agodont check the public ip assignment or if even we are hosting them and shouldnt
Ronnie Sahlberg [Wed, 10 Nov 2010 01:06:05 +0000 (12:06 +1100)]
dont check the public ip assignment or if even we are hosting them and shouldnt
when public ips have been disabled

13 years agoAdd a new tunable : DisableIPFailover that when set to non 0
Ronnie Sahlberg [Tue, 9 Nov 2010 04:19:06 +0000 (15:19 +1100)]
Add a new tunable : DisableIPFailover that when set to non 0
will stopp any ip reallocations at all from happening.

13 years agochange the default for how long to waqit before dropping all ips to 120 seconds
Ronnie Sahlberg [Tue, 9 Nov 2010 01:59:05 +0000 (12:59 +1100)]
change the default for how long to waqit before dropping all ips to 120 seconds

13 years agodont delete all ips from the system during the initial "init" event
Ronnie Sahlberg [Tue, 9 Nov 2010 01:56:02 +0000 (12:56 +1100)]
dont delete all ips from the system during the initial "init" event
leave any ips as they are and let the recovery daemon remove them as required

13 years agowhen creating/adding a public ip, set the initial interface to be the first interface...
Ronnie Sahlberg [Tue, 9 Nov 2010 01:55:20 +0000 (12:55 +1100)]
when creating/adding a public ip, set the initial interface to be the first interface specified

13 years agoBoth nfs and nfslock scripts can fail under redhat in very rare situations.
Ronnie Sahlberg [Thu, 28 Oct 2010 02:43:57 +0000 (13:43 +1100)]
Both nfs and nfslock scripts can fail under redhat in very rare situations.
Ctdb can also be configured to ignore checking for knfsd and if it is alive.
In that situation, no attempt will be made to restart nfs, and sicne nfs is not running,  lockd can not be restarted either.

To workaround this, everytime we try to restart the lockmanager, also try to restart nfsd

13 years agoduring shutdown there is a window after we have stopped TCP and disconnected from...
Ronnie Sahlberg [Thu, 28 Oct 2010 02:38:34 +0000 (13:38 +1100)]
during shutdown there is a window after we have stopped TCP and disconnected from all other nodes but before we have stopped all processing.

During this window we may still hit asynchronous events that will fail because we can not send/receive packets from other nodes.

These messages are logged as ... Transport is DOWN. To help indicate that they are benign messages related to the process of shutting down.

These messages spam the syslog during normal shutdown, so this patch will drop the loglevel of these messages to DEBUG, so that they will not appear in or spam the syslog.

13 years agoremove checking for filesystems and filesystem health from the cnfs script.
Ronnie Sahlberg [Mon, 18 Oct 2010 00:57:38 +0000 (11:57 +1100)]
remove checking for filesystems and filesystem health from the cnfs script.
remove the gpfsmount and gpfsumount entry points

13 years agoIf tdb_open() fails when trying to open the vacuuming database,
Ronnie Sahlberg [Wed, 13 Oct 2010 22:49:23 +0000 (09:49 +1100)]
If tdb_open() fails when trying to open the vacuuming database,
print errno so we get some idea of why this failed.

13 years agotry to restart NFS LOCKD if it failed to start
Ronnie Sahlberg [Wed, 13 Oct 2010 21:12:41 +0000 (08:12 +1100)]
try to restart NFS LOCKD if it failed to start

13 years agoRemove a debug message "Timed out waiting ..."
Ronnie Sahlberg [Tue, 12 Oct 2010 22:21:09 +0000 (09:21 +1100)]
Remove a debug message "Timed out waiting ..."
from the ctdb command.

This is a debugging message and is normal tro tigger on a busy system.
It should not be logged as ERROR.

13 years agoMake sure the statd directory exist before trying to access the
Ronnie Sahlberg [Mon, 11 Oct 2010 21:02:18 +0000 (08:02 +1100)]
Make sure the statd directory exist before trying to access the
"update trigger" file.

CQ 1020344

13 years agomove extracting the config from config.tdb for public addresses
Ronnie Sahlberg [Mon, 11 Oct 2010 15:49:11 +0000 (02:49 +1100)]
move extracting the config from config.tdb for public addresses
into its own function

13 years agoUpdate latency countes to show min/max and average
Ronnie Sahlberg [Mon, 11 Oct 2010 04:11:18 +0000 (15:11 +1100)]
Update latency countes to show min/max and average

13 years agoUpdate the default hash size to be 100001 instead of 10000
Ronnie Sahlberg [Sun, 10 Oct 2010 20:09:18 +0000 (07:09 +1100)]
Update the default hash size to be 100001 instead of 10000
This can sometimes improve performance for environments where very many
files are touched in rapid succession

13 years agodont stop checking interfaces after the first bond device
Ronnie Sahlberg [Fri, 8 Oct 2010 23:54:12 +0000 (10:54 +1100)]
dont stop checking interfaces after the first bond device
continue the loop to process all other interfaces too

13 years agoSpotted by rusty.
Ronnie Sahlberg [Fri, 8 Oct 2010 04:51:44 +0000 (15:51 +1100)]
Spotted by rusty.

Add a missing $
so we delete $_ip   and not _ip

13 years agochange the hash function to use the much better Jenkins hash
Ronnie Sahlberg [Fri, 8 Oct 2010 02:14:14 +0000 (13:14 +1100)]
change the hash function to use the much better Jenkins hash
from the tdb library

cq S1020233

13 years agopytdb: Add __version__ attribute.
Jelmer Vernooij [Mon, 4 Oct 2010 11:17:25 +0000 (13:17 +0200)]
pytdb: Add __version__ attribute.

13 years agopytdb: Include Python.h first to prevent warning.
Jelmer Vernooij [Sat, 2 Oct 2010 21:40:19 +0000 (23:40 +0200)]
pytdb: Include Python.h first to prevent warning.

13 years agopytdb: Check errors after PyObject_New() calls
Kirill Smelkov [Sat, 2 Oct 2010 13:43:50 +0000 (17:43 +0400)]
pytdb: Check errors after PyObject_New() calls

The call could fail with e.g. MemoryError, and we'll dereference NULL
pointer without checking.

Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru>
Signed-off-by: Jelmer Vernooij <jelmer@samba.org>
13 years agopytdb: Add support for tdb_repack()
Kirill Smelkov [Sat, 2 Oct 2010 13:43:46 +0000 (17:43 +0400)]
pytdb: Add support for tdb_repack()

Cc: 597386@bugs.debian.org
Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru>
Signed-off-by: Jelmer Vernooij <jelmer@samba.org>
13 years agopytdb: Add TDB_INCOMPATIBLE_HASH open flag
Kirill Smelkov [Sat, 2 Oct 2010 13:43:40 +0000 (17:43 +0400)]
pytdb: Add TDB_INCOMPATIBLE_HASH open flag

In 2dcf76 Rusty added TDB_INCOMPATIBLE_HASH open flag which selects
Jenkins lookup3 hash for new databases.

Expose this flag to python users too.

Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Jelmer Vernooij <jelmer@samba.org>
13 years agotdb: fix non-WAF build, commit 1.2.6 ABI file.
Rusty Russell [Mon, 27 Sep 2010 01:36:51 +0000 (11:06 +0930)]
tdb: fix non-WAF build, commit 1.2.6 ABI file.

Sorry Jeremy.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
13 years agotdb: TDB_INCOMPATIBLE_HASH, to allow safe changing of default hash.
Rusty Russell [Fri, 24 Sep 2010 06:15:11 +0000 (15:45 +0930)]
tdb: TDB_INCOMPATIBLE_HASH, to allow safe changing of default hash.

This flag to tdb_open/tdb_open_ex effects creation of a new database:
1) Uses the Jenkins lookup3 hash instead of the old gdbm hash if none is
   specified,
2) Places a non-zero field in header->rwlocks, so older versions of TDB will
   refuse to open it.

This means that the caller (ie Samba) can set this flag to safely
change the hash function.  Versions of TDB from this one on will either
use the correct hash or refuse to open (if a different hash is specified).
Older TDB versions will see the nonzero rwlocks field and refuse to open
it under any conditions.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
13 years agotdb: automatically identify Jenkins hash tdbs
Rusty Russell [Fri, 24 Sep 2010 06:09:43 +0000 (15:39 +0930)]
tdb: automatically identify Jenkins hash tdbs

If the caller to tdb_open_ex() doesn't specify a hash, and tdb_old_hash
doesn't match, try tdb_jenkins_hash.

This was Metze's idea: it makes life simpler, especially with the upcoming
TDB_INCOMPATIBLE_HASH flag.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
13 years agotdb: add Bob Jenkins lookup3 hash as helper hash.
Rusty Russell [Fri, 24 Sep 2010 06:04:06 +0000 (15:34 +0930)]
tdb: add Bob Jenkins lookup3 hash as helper hash.

This is a better hash than the default: shipping it with tdb makes it easy
for callers to use it as the hash by passing it to tdb_open_ex().

This version taken from CCAN and modified, which took it from
http://www.burtleburtle.net/bob/c/lookup3.c.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
13 years agotdb: add restore
Volker Lendecke [Sat, 18 Sep 2010 06:56:10 +0000 (10:56 +0400)]
tdb: add restore

Based on an idea by Simon McVittie, largely rewritten

13 years agolib/tdb: fix c++ build warning in tdb_header_hash().
Günther Deschner [Mon, 20 Sep 2010 23:01:51 +0000 (16:01 -0700)]
lib/tdb: fix c++ build warning in tdb_header_hash().

Guenther

13 years agopytdb: Make filename argument optional.
Jelmer Vernooij [Sun, 19 Sep 2010 17:42:29 +0000 (10:42 -0700)]
pytdb: Make filename argument optional.