Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Already on GitHub? Sign in to your account

DHCPD RA Agent #50

Closed
wants to merge 149 commits into
from

Conversation

Projects
None yet

cbowlby commented Jan 27, 2012

This pull request will bring the dhcpd resource agent into the resource-agents repository. I have intentionally kept my changes limited to the one file so that I do not accidentally introduce other issues, as this is my first time within this type of source control. As I get more comfortable working within the github framework, I can work on more elements.

This resource agent script will enable a multi-node (1 active, 1 or more passive) DHCP cluster, adding some fault-tolerance to a critical aspect to network management. While admins most likely use static IP's for their systems, it is sometimes still critical to use a DHCP server, that can become mission critical to maintain.

There is one assumption that is being made at this time and that is that a system admin will want a chrooted DHCP Daemon. Non-chrooted daemons are not yet implemented, though they can be if there are requests.

I will be more then happy to continue to add-value (as time permits), and correct issues (again, as time permits but I will make them priority).

Chris Bowlby added some commits Dec 2, 2011

Chris Bowlby New DHCPD resource agent script
This is the first version of the script and supports very bare bones
starting of the dhcpd daemon. This is to start the review process for
possible contribution into the official linux-ha resource agent package
in the future. Closes #1.
0627b3d
Chris Bowlby Removed bash related code and converted to sh
The script is now SH safe vs being a cross between bash and sh. SH was
chosen over bash, as it appears to be the more selected option.

Closes #3.
1870436
Chris Bowlby Removed old shellfuncs initialization.
The old convention of initializing the shellfunctions from:

: ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/resource.d/heartbeat}
. ${OCF_FUNCTIONS_DIR}/.ocf-shellfuncs

Has been removed and the new convention is being used. Closes #4.
e777660
Chris Bowlby Standardized the PID reference variables.
The OCF_RESKEY_dhcpd_pidfile variable has been renamed to
OCF_RESKEY_dhcpd_pid. Closes #5.
fe6c664
Chris Bowlby Fixed additional variable conventions issues.
The _dhcp portion of the variables has been removed, and a _binary has
been added to the server binary variable itself. Closes #6.
ae397f8
Chris Bowlby The usage function should return success. Closes #7. cdb20ab
Chris Bowlby The default was defined for the required config.
The variable the stores the location of the config file was both
required and defaulted, this should not have been the case and has been
cleaned up. Closes #8.
69f67d6
Chris Bowlby Fixed unique items, and meta data.
The meta data of the variables was not updated when the variable names
themselves was updated, this has been fixed. In addition the config and
chrooted path elements are now tagged as unique, as it might be possible
to run multiple instances under multiple chrooted locations. In
addition, chrooted_path is now flagged as required until such time as
the script is extended to support non-chrooted dhcp servers. Closes #9.
8b3a0f5
Chris Bowlby Remove old test variables for monitoring.
These variables may get reused in the future, but at this time they are
remenants of a failed test to validate the responses from a DHCP server.
Another method needs to be looked into, and for now this will close #10.
00cc581
Chris Bowlby Remove unused dhcpd_getpid function.
This function is a hold-over from the named script I used as a base, and
as such when I reworked the code for the dhcpd application, I ended up
not using this function. Closes #11.
68ae4cd
Chris Bowlby Removed un-needed reload feature.
The reload function simply called a stop/start, and was pointless. For
now I've removed the code and will re-add in the future if there is a
request for it as a feature. Closes #12.
37f9413
Chris Bowlby Added additional checking to dhcpd_stop.
The dhcpd_stop function was not correctly handling returns codes from
dhcpd_monitor. As such, it was "assuming" success, vs actually
having to do other tasks. Additional error checking has been added, and
it may need some more once additional feature support is added.
Closes #13.
f6bb207
Chris Bowlby Remove redundent meta_timewait check.
This check is redundent because the value is always set whenever an
action is involved. Closes #14.
10c44ac
Chris Bowlby Remove un-used chrooted parameter.
This parameter is currently not used as the chrooted_path parameter is
required. As such, the entire configuration is assumed to be chrooted.
Closes #15.
61be6f9
Chris Bowlby Reworked the description of the binary reference.
The binary file that is used to start up dhcpd itself, does not require
the full path.
13063ff
Chris Bowlby Reversed the case for the ocf_is_probe check.
This check was preventing the fail-over of the DHCPD daemon to the
secondary node, as such the check had to be reverse to a NOT probe state
vs a probe state.

Thanks to Andreas Kurz on pointing that out.

Closes #2.
c72bd10
Chris Bowlby Wrap the dhcpd process with the ocf_run helper function.
This function is used to ensure a consistant means of running processes
is available to scripts, as such the dhcpd script should use this. I
have been able to get it working with a suggestions from Dejan
Muhamedagic.

Closes #16.
ad6f6f5
Chris Bowlby Test or [ ] wrappers not neede for ! ocf_is_probe.
Andreas Kurz and Florian Hass have pointed out I do not need to use a
test or [ ] wrapper around the ! ocf_is_probe check. It should be
converted from:

if [ ! ocf_is_probe ] ; then

To:

if !ocf_is_probe; then

and this is done in this commit.

Thanks guys. Closes #17.
8f42f0a
Chris Bowlby Resetting the chrooted_path default parameter.
It has been decided through discussion on the Linux-ha-devel forums to
re-enable the default entry for the chrooted_path variable. The
reasoning is that /var/lib/dhcp is a reasonable default, and it should
not be a required parameter is a default is used.
bc29c6a
Chris Bowlby Updating the copyright year to include 2012 706adff
Chris Bowlby Cleaning up the meta-data descriptions
I have cleaned up the meta-data descriptions to be more descriptive in
what each parameter offers, and how it can be used. Any parameter that
had not-so-obvious side-effects, such as an implied path, have also been
clearly documented.
e8c2e82
Chris Bowlby Merge remote-tracking branch 'upstream/master' 62e2e22

cbowlby commented Feb 27, 2012

As I've not yet done a fork of another project, I've re-synced my tree with the upstream of the cluster-labs tree, I am not sure if that will update this request or not.

Member

fghaas commented Feb 29, 2012

Thanks again for the contribution (and your persistence)!

OK, so the usual instructions pre-merge:

  • Give the RA a final run through ocf-tester.
  • Please rebase your tree against current upstream master.
  • Please edit your commit messages to include severity prefix and the RA name, avoiding -ing forms (example: don't use adding support for superfrobnicate option, instead use Low: dhcpd: support superfrobnicate option)

Both of the latter can be done with an interactive rebase, as explained here.

Thanks!

cbowlby commented Mar 1, 2012

Hi Florian,

I've re-run the ocf-tester command, here is the output from that command:

ocf-tester -n dhcpd -o config="/etc/dhcpd.conf" -o interface="eth0" -o chrooted_path="/var/lib/dhcp" /usr/lib/ocf/resource.d/heartbeat/dhcpd
Beginning tests for /usr/lib/ocf/resource.d/heartbeat/dhcpd...

  • Your agent does not support the notify action (optional)
  • Your agent does not support the demote action (optional)
  • Your agent does not support the promote action (optional)
  • Your agent does not support master/slave (optional)
  • Your agent does not support the reload action (optional)
    /usr/lib/ocf/resource.d/heartbeat/dhcpd passed all tests

I am working on the rebasing as we speak.

Member

fghaas commented Mar 1, 2012

Good; also be sure to run ocf-tester with a limited parameter set, such as only specifying required parameters. And also make sure that it fails when required parameters are not set.

cbowlby commented Mar 1, 2012

Ah, good call, here is the output when specifying the only required parameter:

ocf-tester -n dhcpd -o config="/etc/dhcpd.conf" /usr/lib/ocf/resource.d/heartbeat/dhcpd
Beginning tests for /usr/lib/ocf/resource.d/heartbeat/dhcpd...

  • Your agent does not support the notify action (optional)
  • Your agent does not support the demote action (optional)
  • Your agent does not support the promote action (optional)
  • Your agent does not support master/slave (optional)
  • Your agent does not support the reload action (optional)
    /usr/lib/ocf/resource.d/heartbeat/dhcpd passed all tests

And here is what it does when no parameter is specified:

ocf-tester -n dhcpd /usr/lib/ocf/resource.d/heartbeat/dhcpd
Beginning tests for /usr/lib/ocf/resource.d/heartbeat/dhcpd...

  • rc=7: Start failed. Did you supply enough options with -o ?
    DEBUG: Validating the dhcpd binary exists.
    DEBUG: Validating the chrooted path exists
    DEBUG: Checking to see if the /var/lib/dhcp/ exists and is readable
    DEBUG: Validating the dhcpd user exists
    DEBUG: Validation complete, everything looks good.
    DEBUG: Testing the state of the daemon itself
    INFO: The dhcpd process is not running
    DEBUG: Removing /var/lib/dhcp//var/run/dhcpd.pid if it exists because the DHCPD process is not running.
    DEBUG: Re-validating the exists.
    DEBUG: Creating the /var/lib/dhcp//db/dhcpd.leases file if it does not already exist.
    DEBUG: Removing the old /var/lib/dhcp/dev/urandom device.
    DEBUG: Ensuring all chrooted paths are created, and the required configuration files are copied into place.
    DEBUG: Validating all libraries and sub-dependencies are in place.
    ERROR: Internet Systems Consortium DHCP Server V3.1-ESV Copyright 2004-2010 Internet Systems Consortium. All rights reserved. For info, please visit https://www.isc.org/software/dhcp/ Can't open -chroot: No such file or directory If you did not get this software from ftp.isc.org, please get the latest from ftp.isc.org and install that before requesting help. If you did get this software from ftp.isc.org and have not yet read the README, please read it before requesting help. If you intend to request help from the dhcp-server@isc.org mailing list, please read the section on the README about submitting bug reports and requests for help. Please do not under any circumstances send requests for help directly to the authors of this software - please send them to the appropriate mailing list as described in the README file. exiting.
    DEBUG: Resource Agent Exit Status 7
    DEBUG: dhcpd start returned 7
    Aborting tests

I'm not yet sure why it spits out the ISC's DHCP warning message, though I suspect it is actually coming from the dhcpd process itself, not the script, but it did fail as expected.

akurz and others added some commits Feb 6, 2012

@akurz akurz Med: named: fix bashism - do not rely on $EUID to exist
$EUID is a bashism. It's not available on Debian/Ubuntu where the
default /bin/sh links to /bin/dash.
1ccde80
@akurz akurz Medium: Filesystem: allow to force cloning for local mounts
If you want to run a cluster filesystem like GlusterFS or RADOS/Ceph,
you may also want to have the underlying local filesystem on each node
under cluster control.

For very good reasons local filesystems like ext3/4 or xfs are not
allowed to be mounted on different nodes at the same time, from the
same block device - it may be a shared block device, that allows
concurrent access.

In case of a cluster filesystem setup, where we know for sure that
every node has an exclusive and local block device with a local
filesystem it is valid to allow cloning of such a filesystem
resource. This avoids the need of an extra resource definition
for every node serving parts of a distributed filesystem.

This feature is disabled per default. It is up to the user
to make sure it is safe to enable it for a particular resource!
e37e318
@aspiers aspiers Low: IPaddr2: Fix 'too' typo in description of 'arp_mac' parameter 16aa5a4
@kskmori kskmori Low: IPv6addr: add ocft test case for v3.9.2 3792a0e
@dmuhamedagic dmuhamedagic build: add the IPv6addr ocft test case c7f3afe
@kskmori kskmori Medium: IPv6addr: always use the provided nic and cidr_netmask when s…
…pecified

Related discussion:
  #36
619b31c
@kskmori kskmori Low: IPv6addr: update ocft test case 4623847
@fghaas fghaas Low: doc: update README
Remove the "repository moved" message which is surely confusing
for anyone cloning this repo from the correct GitHub location.

Instead, add a generic description of what this repo really hosts.
e463296

cbowlby commented Mar 1, 2012

Ok, I think the rebase is done, I am attempting to working on the editing my commit comments as soon as I can figure out how to access them. I executed the interactive rebase, however I was not presented any commits of my own, only those from the upstream.

I suspect I'll have to access the commits by directly referencing their guid.

Chris Bowlby and others added some commits Dec 2, 2011

Chris Bowlby Low: dhcpd: New DHCPD resource agent script
This is the first version of the script and supports very bare bones
starting of the dhcpd daemon. This is to start the review process for
possible contribution into the official linux-ha resource agent package
in the future. Closes #1.
147d86b
Chris Bowlby Low: dhcpd: Removed bashisms and converted code to sh.
The script is now SH safe vs being a cross between bash and sh. SH was
chosen over bash, as it appears to be the more selected option.

Closes #3.
571d395
Chris Bowlby Low: dhcpd: Removed old style shellfuncs initialization.
The old convention of initializing the shellfunctions from:

: ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/resource.d/heartbeat}
. ${OCF_FUNCTIONS_DIR}/.ocf-shellfuncs

Has been removed and the new convention is being used. Closes #4.
1ebcf77
Chris Bowlby Low: dhcpd: Standardized the PID reference variables.
The OCF_RESKEY_dhcpd_pidfile variable has been renamed to
OCF_RESKEY_dhcpd_pid. Closes #5.
18b4f6e
Chris Bowlby Low: dhcpd: Fixed additional variable convention issues.
The _dhcp portion of the variables has been removed, and a _binary has
been added to the server binary variable itself. Closes #6.
ef36404
Chris Bowlby Low: dhcpd: The usage function should return success.
Closes #7.
7cf73a0
Chris Bowlby Low: dhcpd: Fixed a default defined for the required config parameter.
The variable the stores the location of the config file was both
required and defaulted, this should not have been the case and has been
cleaned up. Closes #8.
14cfb44
Chris Bowlby Low: dhcpd: Fixed up unique items, and meta data references.
The meta data of the variables was not updated when the variable names
themselves was updated, this has been fixed. In addition the config and
chrooted path elements are now tagged as unique, as it might be possible
to run multiple instances under multiple chrooted locations. In
addition, chrooted_path is now flagged as required until such time as
the script is extended to support non-chrooted dhcp servers. Closes #9.
bf2f6c4
Chris Bowlby Low: dhcpd: Removed old test variables no longer used for monitoring.
These variables may get reused in the future, but at this time they are
remenants of a failed test to validate the responses from a DHCP server.
Another method needs to be looked into, and for now this will close #10.
294aac8
Chris Bowlby Low: dhcpd: Removed unused dhcpd_getpid function.
This function is a hold-over from the named script I used as a base, and
as such when I reworked the code for the dhcpd application, I ended up
not using this function. Closes #11.
07a0b11
Chris Bowlby Low: dhcpd: Removed unused reload option.
The reload function simply called a stop/start, and was pointless. For
now I've removed the code and will re-add in the future if there is a
request for it as a feature. Closes #12.
cb44af0
Chris Bowlby Low: dhcpd: Added additional checking to dhcpd_stop.
The dhcpd_stop function was not correctly handling returns codes from
dhcpd_monitor. As such, it was "assuming" success, vs actually
having to do other tasks. Additional error checking has been added, and
it may need some more once additional feature support is added.
Closes #13.
0d406f0
Chris Bowlby Low: dhcpd: Remove redundant meta_timewait check.
This check is redundent because the value is always set whenever an
action is involved. Closes #14.
40f9d68
Chris Bowlby Low: dhcpd: Remove unused chrooted parameter.
This parameter is currently not used as the chrooted_path parameter is
required. As such, the entire configuration is assumed to be chrooted.
Closes #15.
cad651e
Chris Bowlby Low: dhcpd: Reworked the description of the binary reference.
The binary file that is used to start up dhcpd itself, does not require
the full path.
6582b13
Chris Bowlby Low: dhcpd: Inverted the test case for the ocf_is_probe check.
This check was preventing the fail-over of the DHCPD daemon to the
secondary node, as such the check had to be reverse to a NOT probe state
vs a probe state.

Thanks to Andreas Kurz on pointing that out.

Closes #2.
180fba2
Chris Bowlby Low: dhcpd: Wrap the dhcpd process with the ocf_run helper function.
This function is used to ensure a consistant means of running processes
is available to scripts, as such the dhcpd script should use this. I
have been able to get it working with a suggestions from Dejan
Muhamedagic.

Closes #16.
6a092bf
Chris Bowlby Low: dhcpd: Removed test or [ ] wrappers as they are not needed for !…
… ocf_is_probe.

Andreas Kurz and Florian Hass have pointed out I do not need to use a
test or [ ] wrapper around the ! ocf_is_probe check. It should be
converted from:

if [ ! ocf_is_probe ] ; then

To:

if !ocf_is_probe; then

and this is done in this commit.

Thanks guys. Closes #17.
60d521f
Chris Bowlby Low: dhcpd: Reverting the chrooted_path default parameter change.
It has been decided through discussion on the Linux-ha-devel forums to
re-enable the default entry for the chrooted_path variable. The
reasoning is that /var/lib/dhcp is a reasonable default, and it should
not be a required parameter is a default is used.
f9192ed
Chris Bowlby Low: dhcpd: Updating the copyright year to include 2012. 77757ee
Chris Bowlby Low: dhcpd: Cleaning up meta-data descriptions.
I have cleaned up the meta-data descriptions to be more descriptive in
what each parameter offers, and how it can be used. Any parameter that
had not-so-obvious side-effects, such as an implied path, have also been
clearly documented.
730334c
Léon Keijser High: varnish: new resource agent 168d9d5
Léon Keijser Low: varnish: made changes to conform conventions. Fixed stop routine…
… by adding

an extra check.
0f455d5
Léon Keijser Low: varnish: Better check when stopping varnish. Cosmetic changes. b0531ac
Léon Keijser Low: varnish: Fixed comment. Corrected initialization paths (again). 209baf1
Léon Keijser Low: varnish: Error on child not running. f537bb3
Léon Keijser Low: varnish: Catch exit code of varnishadm instead of matching on ou…
…tput, to

check of varnishd child is running properly.
6a079e7
Léon Keijser Low: varnish: Added configurable admin binary. 4db9c68
Léon Keijser Low: varnish: necessary changes to Makefiles 449595a
@fghaas fghaas Low: man: enclose string parameters in double quotes c74b4bf
Léon Keijser Low: varnish: fix whitespace 075ef46
Léon Keijser Low: varnish: Fixed some cases where ocf_log wouldn't log with the
proper severity level.
fb5033f
Léon Keijser Low: varnish: removed default value for all required variables 452078e
Léon Keijser Low: varnish: reset some default variables where it makes sense to
do so.
4f8bdf4
@fghaas fghaas Low: varnish: remove default for "config"
It's required, so it shouldn't have a default
1e0f4b9
@fghaas fghaas Low: varnish: remove superfluous metadata
If we have no default, we don't need to advertise an empty one.
required="0" and unique="0" are superfluous.

Conflicts:

	heartbeat/varnish
d217e1c
@fghaas fghaas Low: varnish: set OCF_RESOURCE_INSTANCE as default for "instance_name…
…" param

Using the Pacemaker resource ID looks like a sensible default
for the varnish instance name, which will also automatically
provide uniqueness. Allow the user to override this with the
"instance_name" parameter.
63cef14
@fghaas fghaas Low: varnish: shorten "instance_name" to just "name" 9bbc8bf
@fghaas fghaas Low: varnish: remove superfluous metadata
If we have no default, we don't need to advertise an empty one.
required="0" and unique="0" are superfluous.
3ef22e5
Léon Keijser Low: varnish: fixed leftover merge message eaed715
@soushi soushi Fix: DATADIR was being fix value in ocf-tester.in file. 0e84991
@nozawat nozawat Low: slapd: log error if monitor unexpectedly fails
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
3288d0c
@nozawat nozawat Medium: slapd: prevent possible timeout in stop (wrong logic in stop)
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
134ea3e
@nozawat nozawat Low: slapd: fix possible bad validation if the slapd group owner does…
…n't exist

Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
ea57172
@HideoYamauchi HideoYamauchi Low: LVM: add meta-data to the list of supported actions
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
5457fe5
@HideoYamauchi HideoYamauchi Low: LVM: log status to debug only if vgdisplay returned success
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
572cf79
@HideoYamauchi HideoYamauchi Low: LVM: remove obsolete comments
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
37f1ad4
@eotchi eotchi Address syntax error in FAST_STOP calculation d03d290
@kskmori kskmori Medium: IPv6addr: handle a link-local address properly in send_ua ae780b9
@nozawat nozawat Update heartbeat/jboss bdf76c3
AlexanderKrauth Medium : SAPInstance : correcting the unique values of RAs parameters a422da0
AlexanderKrauth Medium: SAPDatabase: correcting the unique values of RAs parameters 6dcfb44
AlexanderKrauth Low: SAPInstance: make sure sidadm is not uninitialized, when calling…
… cleanup_instance
1099cea
Léon Keijser Low: varnish: check for PID file larger than 0 bytes
This fixes an issue where the resource manager would not start
varnishd if a 0 bytes PID file existed. This empty file would get
created if there was a (syntax) error in the configuration file.
d981ff1
@nozawat nozawat Medium: maintenance mode: RA-Status file place in RHEL6 00e538b
@fghaas fghaas Medium: iSCSILogicalUnit: fix default for scsi_sn
Recent versions of openssl prefix the md5 hash with "(stdin)= ",
sed it out if it's present.
d8eeccc
@ikedaj ikedaj Medium: named: make monitor work with IPv6 interfaces
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
88bf2d8
@dmuhamedagic dmuhamedagic Low: Filesystem: add cifs to the list of safe to clone filesystems 4a615f0
@nozawat nozawat Low: rsyslog: test if config file exists
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
f80b071
@nozawat nozawat Medium: slapd: exit with the right code if validation fails (it used …
…to exit always with the generic error code)

Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
08c9808
@nozawat nozawat Low: slapd: add exit code to one error message
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
dd9a763
@nozawat nozawat Low: slapd: improve error reporting in monitor
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
a4375fd
@dmuhamedagic dmuhamedagic Dev: named: update the monitor test regular expression (ammends 778ef5) b15ecc1
@HideoYamauchi HideoYamauchi Low: IPsrcaddr: don't add route if it's already there
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
9caec94
Taro Matsuzawa create RA for Pound 64b2725
@HideoYamauchi HideoYamauchi Low: IPsrcaddr: log an error message if findif fails to find an inter…
…face

Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
59275fa
@dmuhamedagic dmuhamedagic Medium: iscsi: don't fail in stop or monitor if the portal is ambiguo…
…us (bnc#740718)
77cc9a4
@HideoYamauchi HideoYamauchi Low: IPsrcaddr: use ip(8) to find i/f which serves the address
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
3257219
@HideoYamauchi HideoYamauchi Low: IPsrcaddr: don't check if commands we don't use exist
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
a210a3f
@HideoYamauchi HideoYamauchi Low: IPsrcaddr: return early from validate on probes
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
46363c1
@HideoYamauchi HideoYamauchi Low: IPsrcaddr: run validate-all on Linux
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
fdd18e5
@dmuhamedagic dmuhamedagic Low: IPsrcaddr: make a closer match to get the network address in cas…
…e more different networks are configured on the interface
2c005be
@dmuhamedagic dmuhamedagic Low: Tools: ocf-tester: add -q (quiet) option 4b62b1f
@dmuhamedagic dmuhamedagic Dev: Tools: ocf-tester: add debug function to reduce clutter 308b365
@dmuhamedagic dmuhamedagic Low: Tools: ocf-tester: add -d option to debug resource agents df8189a
@dmuhamedagic dmuhamedagic Low: Tools: ocf-tester: update usage and man page 7ed52b6
@nozawat nozawat Medium: apache: add support for IPv6 in monitor
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
82e1a94
@dmuhamedagic dmuhamedagic Low: apache: fix exit code on bad/no configuration b5758f6
@dmuhamedagic dmuhamedagic Low: ocft: apache: update to use default configuration 27b657b
@dmuhamedagic dmuhamedagic Low: ocft: IPsrcaddr: drop lo address, require user to modify the tes…
…t configuration
b4c3fd4
@t-matsuo t-matsuo Medium: ocf-shellfuncs: fix loglevel variable scope in ha_log
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
5e7ec4c
@nozawat nozawat Low: tomcat: use the PID file to check if the process is running
pgrep(1) uses extended RE, so searching for an arbitrary string
("\\""${CATALINA_OPTS}") may produce wrong results.

Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
beeb59a
@nozawat nozawat Low: tomcat: fix a comment
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
af8fc9d
@nozawat nozawat Low: tomcat: implement the status action properly
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
0905ae8
Kazuyoshi Mii Medium: lxc: fix LXC_status to work with lxc-0.7.5 or later 1f2c436
@nozawat nozawat Low: tomcat: add basic validation
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
fd898a8
@lge lge Medium: fix exportfs monitor action for special characters and common…
… suffixes

Trying to match a multiline blob against an unquoted single line literal
string, interpreted as unanchored regular expression, does not work.

The directory name (or more likely, client spec) may contain regex
"special" characters, such as a '*' wildcard, which caused the previous
monitor action implementation to fail.

Also, if /foo/bar was exported, but /bar was not,
still the monitoring for /bar would return "running"
(if exported to a similar enough client spec).

Fixed by unwrapping the output of exportfs,
then doing a line-wise literal match.
f42df95
@fabbione fabbione fix resource-agents.spec.in %post section
Signed-off-by: Fabio M. Di Nitto <fdinitto@redhat.com>
1d2108f
@fabbione fabbione spec: silence cp error in case there is empty directory
patch from Tomo <nozawat@gmail.com>

Signed-off-by: Fabio M. Di Nitto <fdinitto@redhat.com>
c1c9eed
@nozawat nozawat Low: slapd: update ocf library path
Signed-off-by: Dejan Muhamedagic <dejan@suse.de>
9cd904c
@tserong tserong Medium: Tools: findif: Use most specific matching route (bnc#740738)
See #52 for
discussion.  Thanks to Timur Bakeyev for original diagnosis and
patch.

Signed-off-by: Tim Serong <tserong@suse.com>
fafac01
Kazuyoshi Mii Medium: enhancement of ocf_version_cmp to compare version of differen…
…t scheme correctly, such as 0.7.4.2 and 0.7.5
0054420
@dmuhamedagic dmuhamedagic Medium: oralsnr: match only one (our) listener process
In case there are multiple oracle instances and multiple tnslsnr
processed we'd match all of them. On stop, this could result in
stopping other listeners.

Thanks to Alexander Milcz.
0602918
@AlexanderKrauth AlexanderKrauth Low: SAPDatabase: make use of ocf_is_root f86cd6b
@AlexanderKrauth AlexanderKrauth Medium: SAPDatabase: switch to status method in case of probe 261c359
@AlexanderKrauth AlexanderKrauth Medium: SAPDatabase: downward compatible functions into a separate sc…
…ript file - sapdb-nosha.sh
500bddb
@AlexanderKrauth AlexanderKrauth High: SAPDatabase: version 2.00 make use of saphostagent 81b4c72
@AlexanderKrauth AlexanderKrauth High: SAPDatabase: add support for Sybase ASE and SAP HANA database bca8837
@dmuhamedagic dmuhamedagic build: add the new RA pound ac729a0
@AlexanderKrauth AlexanderKrauth Low: SAPDatabase: use default format specifier of ps command 2842912
@esimone74 esimone74 Medium: Xen: add support for HVM ACPI graceful shutdown 0d44ad7
@dmuhamedagic dmuhamedagic Medium: VirtualDomain: if the configuration file is missing on stop e…
…xit with success
55bfdd4
@akurz akurz Med: named: fix bashism - do not rely on $EUID to exist
$EUID is a bashism. It's not available on Debian/Ubuntu where the
default /bin/sh links to /bin/dash.
b01d8f0
@fghaas fghaas Low: Xen: fix longdesc for shutdown_acpi 766375e
@lge lge Medium: Filesystem: improve read/write checks for CHECK_LEVEL 10, 20
Some devices don't support 512 byte direct writes anymore,
the "physical" block size is 4kByte.

Changing "bs=512 count=16" to "bs=4k count=1" also helps to avoid
spurious timeouts.  I've seen badly tuned (and/or totally overloaded) IO
stacks exposing a synchronous single-request IO latency of many seconds
(under load).  If we need 16 of those, we can easily violate timeout
settings.
96ac5f6
@dmuhamedagic dmuhamedagic Low: Filesystem: remove loading the scsi_hostadapter kernel module b410465
@dmuhamedagic dmuhamedagic Low: Raid1: remove loading the scsi_hostadapter kernel module 38ea008
@fghaas fghaas Medium: asterisk: don't try to shut down with "core stop now"
If an asterisk instance is thoroughly shot, the "core stop now"
command will block. Since we'd been doing this in the foreground,
this would mean we could run into a timeout and an ugly stop failure.

In addition, doing "core stop now" adds no real benefit over just
sending a SIGTERM, where sending the SIGTERM is asynchronous and
hence, will not block -- thus our safeguards have a chance to kick
in before the timeout expires, and we can actually avoid the stop
failure.

Thanks to Rob Thomas for pointing this out.
f842707
@akurz akurz Medium: Filesystem: allow to force cloning for local mounts
If you want to run a cluster filesystem like GlusterFS or RADOS/Ceph,
you may also want to have the underlying local filesystem on each node
under cluster control.

For very good reasons local filesystems like ext3/4 or xfs are not
allowed to be mounted on different nodes at the same time, from the
same block device - it may be a shared block device, that allows
concurrent access.

In case of a cluster filesystem setup, where we know for sure that
every node has an exclusive and local block device with a local
filesystem it is valid to allow cloning of such a filesystem
resource. This avoids the need of an extra resource definition
for every node serving parts of a distributed filesystem.

This feature is disabled per default. It is up to the user
to make sure it is safe to enable it for a particular resource!
cf086cb
@fghaas fghaas Low: iSCSITarget: fix typo in metadata
Remove trailing whitespace from declaration of "status" operation.
Thanks to Rasto Levrinc for spotting the problem.
6ac172e
@fghaas fghaas Low: iSCSITarget: fix C&P error in metadata
Make shortdesc for "implementation" parameter more meaningful.
Thanks to Rasto Levrinc for the report.
333f69f
@aspiers aspiers Low: IPaddr2: Fix 'too' typo in description of 'arp_mac' parameter 1cdfdf3
@kskmori kskmori Low: IPv6addr: add ocft test case for v3.9.2 17894c4
@dmuhamedagic dmuhamedagic build: add the IPv6addr ocft test case a480b1b
@fghaas fghaas Low: doc: update README
Remove the "repository moved" message which is surely confusing
for anyone cloning this repo from the correct GitHub location.

Instead, add a generic description of what this repo really hosts.
278afa0
@kskmori kskmori Medium: IPv6addr: always use the provided nic and cidr_netmask when s…
…pecified

Related discussion:
  #36
ff90b76
@kskmori kskmori Low: IPv6addr: update ocft test case c7c77de
Chris Bowlby Low: varnish: fixing a merge issue 2e68dc0

cbowlby commented Mar 1, 2012

Ok, I've been able to rebase all of my commits, and have also been able to push them all back into my repo, it has bloated this pull request with all of those commits, so I hope I did this right.. lol

Contributor

dmuhamedagic commented Apr 15, 2012

On the SLE11SP2 there is a problem with /proc/net/dev not being available because proc is not mounted in chroot. The error text:

Error opening '/proc/net/dev' to list interfaces
Can't get list of interfaces.

cbowlby commented Apr 15, 2012

Hi Dejan,

That's most likely due to the v3 kernel implementation on SP2, as soon
as I have a moment I'll get something built for testing to see what
changes I'll need to make, thanks.

On 04/15/2012 04:11 PM, Dejan Muhamedagic wrote:

On the SLE11SP2 there is a problem with /proc/net/dev not being available because proc is not mounted in chroot. The error text:

Error opening '/proc/net/dev' to list interfaces
Can't get list of interfaces.


Reply to this email directly or view it on GitHub:
#50 (comment)

Contributor

dmuhamedagic commented Apr 16, 2012

On Sun, Apr 15, 2012 at 03:22:17PM -0700, cbowlby wrote:

That's most likely due to the v3 kernel implementation on SP2

It's nothing to do with the kernel. What's missing is the proc
mount in chroot. It probably has to do with the dhcpd version
(4.2.3 in this case).

cbowlby commented May 15, 2012

I am going to re-fork my repo to fix the issues with all these patches. After a quick discussion with Dejan it was determined the history is not important for a new resource. I'll simply import my finalized DHCP script into the re-forked instance, and submit another pull request once done.

@cbowlby cbowlby closed this May 15, 2012

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment