fixup tests/dbcheck: One way links are expected to be stale

Signed-off-by: Andrew Bartlett <abartlet@samba.org>

renamedc: Make a more targeted dbcheck

Signed-off-by: Garming Sam <garming@catalyst.net.nz>

tests/dbcheck: One way links are expected to be stale

Run a targeted dbcheck to fix only the one way links.

Signed-off-by: Garming Sam <garming@catalyst.net.nz>

dbcheck: fix a typo

Signed-off-by: Garming Sam <garming@catalyst.net.nz>

dbcheck: fix rules argument to specify a partial --yes

Signed-off-by: Garming Sam <garming@catalyst.net.nz>

Amend test

dbcheck linked attribute tests: save environment with bad links

We save a database snapshot that contains linked attributes that
should have been deleted, and make sure dbcheck fixes those links
without ruining anything else.

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>

blackbox/dbcheck-oldrelease: more accurate temp filename

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>

s4/selftest/provisions/dump.sh: dump to target dir if supplied

This is clearly what was meant to happen.

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>

dsdb tests: add linked attribute tests

Note that this test will not work properly across ldap as the
marked-deleted linked attributes will not appear.

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>

dbcheck: check for linked atributes that should not exist

In order to do this we need to use the reveal internals control, which
breaks the comparison against extended DNs. So we compare the
components instead.

Because this patch makes our code notice and fix stale one-way-links
(eg, after a rename) now, the renamedc test needs to be adjusted to
match.

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>

dbcheck: cache linkIDs and reverse attribute names

This avoids fetching the same same schema things again and again.

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>

dsdb: add vanish links control

Normally linked attributes are deleted by marking them as with RMD flags,
but sometimes we want them to vanish without trace. At those times we
set the DSDB_CONTROL_REPLMD_VANISH_LINKS control.

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>
Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Signed-off-by: Bob Campbell <bobcampbell@catalyst.net.nz>
Pair-programmed-with: Andrew Bartlett <abartlet@samba.org>

repl_meta_data: free context on error in replmd_modify_la_delete()

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>

replmd_modify_delete: check talloc_new()

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>

s4/dsdb/repl_meta_data: use local bool version of flag

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>

tests/dsdb/sam.py: remove duplicated primaryGroupID change

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>

Remove unused stf directory

Signed-off-by: Douglas Bagnall <douglas.bagnall@catalyst.net.nz>

s4-torture: Add AES and RC4 enctype checks

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Andrew Bartlet <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Wed Jul  6 19:06:19 CEST 2016 on sn-devel-144

s4-torture: Add torture_check_krb5_error() function

Signed-off-by: Andreas Schneider <asn@samba.org>
Reviewed-by: Andrew Bartlet <abartlet@samba.org>

schema: Reorder dsdb_set_schema() to unlink the old schema last

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>

dsdb: Remove 120 second delay and USN from schema refresh check

We now refresh it once the schema changes, so that replication can
proceed right away.  We use the sequence number in the metadata.tdb.

The previous commit added a cache for this value, protected by
tdb_seqnum().

metadata.tdb is now opened at startup to provide this support.

Note that while still supported, schemaUpdateNow is essentially rudundent:
instead, to ensure we increment the sequence number correctly, we unify that check
into repl_meta_data at the transaction close.

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>

dsdb: Remove use of schema USN in samldb_add_handle_msDS_IntId

This is not a frequent enough operation to warrent a cache, and the USN will be removed
from the schema code shortly

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>

schema: Make the fetch of the schema version fast

Use the tdb_seqnum() to avoid needing locks to check if the schema has not changed

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andreas Schneider <asn@samba.org>

ldb: Avoid use-after-free when one error message is printed into another

Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andreas Schneider <asn@samba.org>

provision: Ignore duplicate attid and governsID check

During the provision this causes a huge performance hit as these two
attributes are unindexed.

Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Signed-off-by: Bob Campbell <bobcampbell@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Pair-programmed-with: Garming Sam <garming@catalyst.net.nz>

provision_fill: move GPO into transaction

Signed-off-by: Bob Campbell <bobcampbell@catalyst.net.nz>
Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Pair-programmed-with: Garming Sam <garming@catalyst.net.nz>

provision_fill: move most db accesses into transactions

Signed-off-by: Bob Campbell <bobcampbell@catalyst.net.nz>
Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Pair-programmed-with: Garming Sam <garming@catalyst.net.nz>

ctdb-scripts: Quote some variable expansions

This avoids relevant shellcheck warnings.  This is most of the
shellcheck low hanging fruit in the non-test code.  Many of the other
warnings produced by shellcheck are either false positives, are
non-trivial to fix or a fix may result in worse code.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Amitay Isaacs <amitay@samba.org>
Autobuild-Date(master): Wed Jul  6 08:15:49 CEST 2016 on sn-devel-144

ctdb-scripts: Fix incorrect variable reference

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-scripts: Use globs instead of ls to list files

shellcheck reports that using ls is fragile.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-scripts: Fix incorrect variable reference

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-scripts: Fix incorrect variable reference

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-scripts: Update script boilerplate to avoid shellcheck warnings

* Assign the output of dirname to temporary variable to avoid word
  splitting when directory name contains whitespace

* Drop export of CTDB_BASE to avoid masking broken return value -
  functions file does the export anyway

* Quote path when including functions file

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-scripts: Export CTDB_BASE in functions file

This avoids having to export it in every file that includes the
functions file.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-scripts: Drop optional argument to nfs_check_services()

Added so that nfs_check_services() could be run against an arbirary
directory.  However, with the function moved to the event script, this
isn't useful.  CTDB_NFS_CHECKS_DIR can be used for testing instead.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

gensec: Change log level for message when obtaining PAC from gss_get_name_attribute failed

This is the second part for the issue from commit 8bb4fccd. A KDC that
does not return a PAC first triggers this message, then the "resorting
to local user lookup" one. Change the log level for the "obtaining PAC
via GSSAPI gss_get_name_attribute" message as well to avoid spamming the
logs during normal usage. While changing this message, also remove the
discard_const since it is no longer required.

Signed-off-by: Christof Schmitt <cs@samba.org>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Wed Jul  6 04:27:03 CEST 2016 on sn-devel-144

auth: fix a memory leak in gssapi_get_session_key()

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12006

Signed-off-by: Uri Simchoni <uri@samba.org>
Signed-off-by: Richard Sharpe <rsharpe@samba.org>
Signed-off-by: Andreas Schneider <asn@samba.org>
Autobuild-User(master): Richard Sharpe <sharpe@samba.org>
Autobuild-Date(master): Wed Jul  6 00:40:15 CEST 2016 on sn-devel-144

s3-libads: fix a memory leak in ads_sasl_spnego_bind()

BUG: https://bugzilla.samba.org/show_bug.cgi?id=12006

Signed-off-by: Uri Simchoni <uri@samba.org>
Signed-off-by: Richard Sharpe <rsharpe@samba.org>
Signed-off-by: Andreas Schneider <asn@samba.org>

ctdb-build: Exit if requested feature cannot be built

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>
Autobuild-User(master): Martin Schwenke <martins@samba.org>
Autobuild-Date(master): Tue Jul  5 14:38:30 CEST 2016 on sn-devel-144

ctdb-daemon: Log ctdb socket in the main daemon

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-pmda: CTDB client code does not require ctdb->methods

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-daemon: Check if method is initialized before calling

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-ib: Include system/wait.h for signal

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Expose ctdb_ltdb_fetch in client API

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Add debug messages to client db api

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Fix implementation of transaction cancel

Wrap async transaction cancel to unlock g_lock lock and free transaction
handle.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Add async version of transaction cancel

Transaction cancel should get rid of g_lock lock.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Fix implementation of transaction commit

There is no need to explicitly check that recovery is not active before
sending TRANS33_COMMIT control.  Just try TRANS3_COMMIT control and if
recovery occurs before the control is completed, the control will fail
and it can be retried.

Make sure g_lock lock is released after the transaction is complete.
Also, add timeout to the client api.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Fix implementation of transaction start

Since g_lock checks if the process exists in case of conflicting lock,
there is no need to register srvid.

Transaction start returns a transaction handle and transaction
commit/cancel will free that handle.  Since we cannot call async code
in a talloc destructor, this avoids the use of talloc destructor for
cancelling the transaction.

If user frees the transaction handle instead of calling transaction
cancel, it will leave stale g_lock lock.  This stale g_lock lock  will
get cleaned up on next transaction attempt.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: During transaction commit fetch seqnum locally

This avoids extra controls to the server.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Release the g_lock record once the update is done

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Remove commented old g_lock implemention code

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Release g_lock lock before retrying

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Fix g_lock implementation

If a conflicting g_lock entry is found, check if the process exists.
This matches Samba implementation.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: If g_lock lock conflicts, try again sooner

Instead of delaying for 1 second, try to get g_lock lock again after
1 milli-second.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Factor out ctdb_client_get_server_id function

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Use async version of delete_record in g_lock unlock

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Fix implementation of delete_record

In delete_record, sync call to ctdb_ctrl_schedule_for_deletion will
cause nested event loops.  Instead wrap the async version.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Add async version of delete_record

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Fix ctdb_rec_buffer traversal routine

In commit 1ee7053180057ea526870182b5619a206b4d103b, the
ctdb_rec_buffer_traverse always passes NULL for header.  So explicitly
extract header from the data.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-client: Add sync version of sending multiple messages

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-daemon: Improve log message

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

ctdb-recoverd: Improve election win messages

Logging that node has lost election is less useful than knowing which
node has won the election.

Signed-off-by: Amitay Isaacs <amitay@gmail.com>
Reviewed-by: Martin Schwenke <martin@meltin.net>

kerberos: Return enc data on PREAUTH_FAILED

Without the enc data, Windows clients will perform two AS-REQ causing the password
lockout count to increase by two instead of one.

Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
BUG: https://bugzilla.samba.org/show_bug.cgi?id=11539

Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Tue Jul  5 10:52:32 CEST 2016 on sn-devel-144

password_hash: Make an error message clearer

Signed-off-by: Bob Campbell <bobcampbell@catalyst.net.nz>
Reviewed-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>
Autobuild-User(master): Garming Sam <garming@samba.org>
Autobuild-Date(master): Tue Jul  5 03:47:52 CEST 2016 on sn-devel-144

typo: mandetory -> mandatory

Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

typo: componemt => component

Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

tevent: typo in documentation

Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

param: fix a typo emtpy -> empty

Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

check_password_script: Add a DEBUG message for timeouts

Signed-off-by: Bob Campbell <bobcampbell@catalyst.net.nz>
Pair-programmed-with: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

selftest: add check password script test

Pair-programmed-with: Garming Sam <garming@catalyst.net.nz>
Signed-off-by: Bob Campbell <bobcampbell@catalyst.net.nz>
Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

check-password-script: Allow AD to execute these scripts

In contrast to source3, this is run as root and without substitution.

Signed-off-by: Garming Sam <garming@catalyst.net.nz>
Reviewed-by: Andrew Bartlett <abartlet@samba.org>

ctdb-tests: Link to ctdb-ipalloc instead of using ctdbd_test.c

Less code, quicker build time, smaller binary...

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>
Autobuild-User(master): Amitay Isaacs <amitay@samba.org>
Autobuild-Date(master): Mon Jul  4 19:29:08 CEST 2016 on sn-devel-144

ctdb-ipalloc: Drop implicit dependency on ctdb-common

Use new functions from protocol API instead.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-tests: Allow takeover tests to be run under valgrind

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-tests: Drop use of CTDB context from takeover test

The ipalloc code doesn't need a CTDB context so neither should the
code that tests it.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: IP allocation state is now an opaque structure

It is private to the IP allocation module.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: ipalloc() returns public IP list

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Move set_ipflags_internal() to ipalloc

Rename it ipalloc_set_node_flags().

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Switch set_ipflags_internal() to use a new-style node map

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Move ipalloc state initialisation to ipalloc.c

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Pass extra data to IP allocation state initialisation

No longer require CTDB context but pass in number of nodes, algorithm,
no_ip_failback and force_rebalance_nodes.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Make no_ip_failback a boolean

No need to expose tunable values that far down.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Fix buggy short-circuit when no IPs are available

At the moment IP is short-circuited when there are no available IP
addresses.  However, if some IP addresses are already allocated then
"no available IP addresses" means that all the addresses should
(probably) be released.  The current short-circuit means that no
already hosted IP addresses will be released.

The short-circuit exists to avoid lots of messages saying that all IP
addresses can not be assigned at startup time.  So, add a check to
ipalloc_can_host_ips() so that it succeeds if IP addresses are already
allocated to nodes.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: New function ipalloc_can_host_ips()

Abstracts out code involving internals of IP allocation state.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Drop known public IPs from IP allocation state

This is never used in the allocation algorithms.  It is only used when
building the merged IP list.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Move create_merged_ip_list() into ipalloc

How the existing IP layout is constructed and how the merged IP list is
sorted are important aspects of the IP allocation algorithm.  Construct the
merged IP list when known and available IPs are assigned.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: New function ipalloc_set_public_ips()

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Remove function ctdb_reload_remote_public_ips()

Use ctdb_fetch_remote_public_ips() inline to fetch each list.  Assign
them into the IP allocation state separately.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Clean up reloading of remote public IPs

Factor out new function ctdb_fetch_remote_public_ips() to fetch known
or available public IP addresses, according to flags.

This also drops the hack where the array from a
ctdb_public_ip_list_old was assigned to a pointer in a
ctdb_public_ip_list.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Don't build a global IP tree

It isn't used outside this function, so just use a local variable.

This makes create_merged_ip_list() independent of the CTDB context.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Drop code to update IP assignment tree

This code is not used.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-tools: Don't bother sending CTDB_SRVID_RECD_UPDATE_IP

Nothing is listening.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-recoverd: Drop code to change the IP assignment tree

The tree is no longer used in verification.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Drop remote IP verification

It is only run during a takeover run and only logs errors.  It doesn't
actually do anything to fix potential errors.  The takeover run should
fix any inconsistencies anyway.

Instead, leave a comment in the recovery daemon's monitoring loop to
add proper remote IP verification later.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Drop a use of CTDB_NO_MEMORY_NULL()

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Do not use node count or PNNs from CTDB context

This is unnecessary.  IP allocation state already has a node count and
"i" is already a PNN.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Drop an unnecessary check

Deleted (and other inactive) nodes will have an empty list of known
IP addresses.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>

ctdb-ipalloc: Move if-statement with broken condition

This pointer is for an array that is always allocated.  The check is
meant to skip a node that has no IP addresses.  However, when there
are no IP addresses the loop below will not do anything anyway.

Add this as a check at the beginning of the function instead.

Signed-off-by: Martin Schwenke <martin@meltin.net>
Reviewed-by: Amitay Isaacs <amitay@gmail.com>