dejanmm at fastmail
Jun 22, 2010, 5:07 AM
Post #1 of 1
(488 views)
Permalink
|
|
Re: [Pacemaker] pgpool2 OCF Resource Agent
|
|
Hi,
On Mon, Jun 21, 2010 at 04:53:52PM -0400, Eliot Gable wrote:
> Here is a pgpool2 resource agent. I'm not sure where to submit
> this for inclusion into the project, so I'm sending it here. If
> I need to post it somewhere else, let me know.
linux-ha-dev is the right place. You'll have to subscribe to
post. I'll move the discussion there. Also, please attach
scripts instead of pasting them.
Did you run ocf-tester on this RA?
> #!/bin/sh
> #
> # pgpool-II resource agent.
> #
> # This program is free software; you can redistribute it and/or modify
> # it under the terms of version 2 of the GNU General Public License as
> # published by the Free Software Foundation.
> #
> # This program is distributed in the hope that it would be useful, but
> # WITHOUT ANY WARRANTY; without even the implied warranty of
> # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
> #
> # Further, this software is distributed without any warranty that it is
> # free of the rightful claim of any third person regarding infringement
> # or the like. Any license provided herein, whether implied or
> # otherwise, applies only to this software file. Patent licenses, if
> # any, provided herein do not apply to combinations of this program with
> # other software, or any other product whatsoever.
> #
> # You should have received a copy of the GNU General Public License
> # along with this program; if not, write the Free Software Foundation,
> # Inc., 59 Temple Place - Suite 330, Boston MA 02111-1307, USA.
> #
> #######################################################################
> #
> # This resource agent was written by Eliot Gable <egable [at] gmail>
> #
> #######################################################################
>
> #######################################################################
> # Initialization:
>
> : ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/resource.d/heartbeat}
> . ${OCF_FUNCTIONS_DIR}/.ocf-shellfuncs
>
> #######################################################################
>
> meta_data() {
> cat <<END
> <?xml version="1.0"?>
> <!DOCTYPE resource-agent SYSTEM "ra-api-1.dtd">
> <resource-agent name="pgpool2" version="0.1">
> <version>1.0</version>
>
> <longdesc lang="en">
> This resource agent provides basic management of pgpool-II.
> It starts and stops pgpool-II and monitors its status. It will
> also monitor the status of each connection and can optionally
> attempt to automatically reconnect detached nodes or can
> mark the service as failed if there are any detached nodes.
> </longdesc>
> <shortdesc lang="en">Manages pgpool-II</shortdesc>
>
> <parameters>
>
> <parameter name="pcp_admin_username" required="1">
> <longdesc lang="en">
> This specifies the administrative username for pgpool-II control.
> </longdesc>
> <shortdesc lang="en">Administrative username.</shortdesc>
> <content type="string" default="" />
> </parameter>
It would be good to reword all longdesc for parameters. "This
specifies" is superfluous.
> <parameter name="pcp_admin_password" required="1">
> <longdesc lang="en">
> This specifies the administrative password for pgpool-II control.
> </longdesc>
> <shortdesc lang="en">Administrative password.</shortdesc>
> <content type="string" default="" />
> </parameter>
>
> <parameter name="pcp_admin_port" required="1">
> <longdesc lang="en">
> This specifies the administrative port for pgpool-II control.
> </longdesc>
> <shortdesc lang="en">Administrative port for PCP commands.</shortdesc>
> <content type="string" default="" />
> </parameter>
>
> <parameter name="pcp_admin_host">
> <longdesc lang="en">
> This specifies the administrative host for pgpool-II control.
> </longdesc>
> <shortdesc lang="en">Administrative host for PCP commands.</shortdesc>
> <content type="string" default="localhost" />
> </parameter>
The defaults are not set automatically. The script has to do that
itself:
OCF_RESKEY_pcp_admin_host_default="localhost"
...
: ${OCF_RESKEY_pcp_admin_host=$OCF_RESKEY_pcp_admin_host_default}
> <parameter name="pgpool_bin">
> <longdesc lang="en">
> This parameter specifies the path to the pgpool-II binary.
> </longdesc>
> <shortdesc lang="en">Path to pgpool.</shortdesc>
> <content type="string" default="/usr/bin/pgpool" />
> </parameter>
>
> <parameter name="pcp_attach_node_bin">
> <longdesc lang="en">
> This parameter specifies the path to the pcp_attach_node binary.
> </longdesc>
> <shortdesc lang="en">Path to pcp_attach_node.</shortdesc>
> <content type="string" default="/usr/bin/pcp_attach_node" />
> </parameter>
>
> <parameter name="pcp_detach_node_bin">
> <longdesc lang="en">
> This parameter specifies the path to the pcp_detach_node binary.
> </longdesc>
> <shortdesc lang="en">Path to pcp_detach_node.</shortdesc>
> <content type="string" default="/usr/bin/pcp_detach_node" />
> </parameter>
>
> <parameter name="pcp_node_count_bin">
> <longdesc lang="en">
> This parameter specifies the path to the pcp_node_count binary.
> </longdesc>
> <shortdesc lang="en">Path to pcp_node_count.</shortdesc>
> <content type="string" default="/usr/bin/pcp_node_count" />
> </parameter>
>
> <parameter name="pcp_node_info_bin">
> <longdesc lang="en">
> This parameter specifies the path to the pcp_node_info binary.
> </longdesc>
> <shortdesc lang="en">Path to pcp_node_info.</shortdesc>
> <content type="string" default="/usr/bin/pcp_node_info" />
> </parameter>
>
> <parameter name="stop_mode">
> <longdesc lang="en">
> This parameter specifies the stop mode to use when stopping pgpool-II.
> </longdesc>
> <shortdesc lang="en">Stop mode for pgpool-II.</shortdesc>
> <content type="string" default="f" />
> </parameter>
>
> <parameter name="auto_reconnect">
> <longdesc lang="en">
> If this parameter is set to "true", then during monitoring actions,
> the resource agent will attempt to re-attach any disconnected
> nodes. No error will be reported if re-attachment fails.
> </longdesc>
> <shortdesc lang="en">Automatically reattach failed nodes</shortdesc>
> <content type="string" default="" />
This should be content type="boolean".
> </parameter>
>
> <parameter name="fail_on_detached">
> <longdesc lang="en">
> This parameter instructs the resource agent to mark pgpool-II in a
> failed state if one or more of the nodes is detached. The monitor
> action will always mark pgpool-II in a failed state if all nodes are
> detached, so this is only useful if you want to mark pgpool-II in a
> failed state if at least one node is detached. The auto_reconnect
> option will always try to reconnect detached nodes (if enabled)
> before this fail_on_detached mechanism triggers.
> </longdesc>
> <shortdesc lang="en">Marks resource as failed if at least one node is detached.</shortdesc>
> <content type="string" default="" />
> </parameter>
>
> <parameter name="fail_on_node_detached">
> <longdesc lang="en">
> This parameter is similar to fail_on_detached, except you can
> specify a comma-seperated list of node IDs. If specified, pgpool2
> will only be marked as "failed" if one of the nodes in the list
> is detached or if all nodes are detached.
> </longdesc>
> <shortdesc lang="en">Specify a list of nodes to monitor for failure.</shortdesc>
> <content type="string" default="" />
> </parameter>
>
> </parameters>
>
> <actions>
> <action name="start" timeout="20" />
> <action name="stop" timeout="40" />
> <action name="monitor" timeout="20" interval="5"
> depth="0"/>
> <action name="reload" timeout="20" />
I suppose that these timeouts are the minimum for some standard
deployment.
> <action name="meta-data" timeout="5" />
> <action name="validate-all" timeout="20" />
> </actions>
> </resource-agent>
> END
> }
>
> #######################################################################
>
> pgpool2_usage() {
> cat <<END
> usage: $0 {start|stop|status|monitor|validate-all|meta-data}
>
> Expects to have a fully populated OCF RA-compliant environment set.
> END
> }
>
> pgpool2_start() {
> pgpool2_status
> status=$?
It would be good to declare variables as local.
But here you can get by without one:
if pgpool2_status; then
...
fi
> if [ $status -eq $OCF_SUCCESS ]; then
> ocf_log debug "${OCF_RESOURCE_INSTANCE} $__OCF_ACTION : already started."
> return $OCF_SUCCESS
> fi
> if $PGPOOL; then
> sleep 2
> pgpool2_status
> status=$?
> if [ $status -eq $OCF_SUCCESS ]; then
> ocf_log info "${OCF_RESOURCE_INSTANCE} Successfully started pgpool-II"
> return $OCF_SUCCESS
> else
> ocf_log error "${OCF_RESOURCE_INSTANCE} Failed to start pgpool-II"
> return $OCF_ERR_GENERIC
> fi
> else
> ocf_log error "${OCF_RESOURCE_INSTANCE} Failed to start pgpool-II"
> fi
Perhaps it would be good to add the exit code into the error log
messages. If that could help in troubleshooting.
> return $OCF_ERR_GENERIC
> }
>
> pgpool2_stop() {
> pgpool2_status
> status=$?
> case $status in
> $OCF_SUCCESS)
> ocf_log info "Using $PGPOOL -m $STOP_MODE stop to stop pgpool-II"
> if $PGPOOL -m $STOP_MODE stop; then
> ocf_log info "${OCF_RESOURCE_INSTANCE} Successfully stopped pgpool-II"
> return $OCF_SUCCESS
> else
> ocf_log error "${OCF_RESOURCE_INSTANCE} Failed to stop pgpool-II"
> fi
> ;;
> $OCF_NOT_RUNNING)
> ocf_log debug "${OCF_RESOURCE_INSTANCE} $__OCF_ACTION : already stopped."
> return $OCF_SUCCESS
> ;;
> esac
> return $OCF_ERR_GENERIC
> }
>
> pgpool2_status() {
> if [ ! -r "/var/run/pgpool/pgpool.pid" ]; then
> return $OCF_NOT_RUNNING
> fi
> ps_info=$(ps ax | grep pgpool | grep -v pgpool: | grep -v grep | grep $(cat /var/run/pgpool/pgpool.pid))
Don't like the way this grep looks, perhaps better sth like this:
ps_info=$(ps ax | grep "[p]gpool[^:]" | grep $(cat /var/run/pgpool/pgpool.pid))
> if [ -z "${ps_info}" ]; then
> return $OCF_NOT_RUNNING
> else
> # Try to reconnect any detached nodes
> if [ "${OCF_RESKEY_auto_reconnect}" == "true" ] ||
> [ "${OCF_RESKEY_auto_reconnect}" == "t" ] ||
> [ "${OCF_RESKEY_auto_reconnect}" == "yes" ] ||
> [ "${OCF_RESKEY_auto_reconnect}" == "y" ]; then
Use if is_ocf_true ${OCF_RESKEY_auto_reconnect}
> NODE_COUNT=$($PCP_NODE_COUNT 1 $OCF_RESKEY_pcp_admin_host $OCF_RESKEY_pcp_admin_port $OCF_RESKEY_pcp_admin_username $OCF_RESKEY_pcp_admin_password)
> for (( node=0; $node < $NODE_COUNT; node=(($node+1)) )); do
I'd rather not have fancy bash constructs. How about:
for node in `seq 0 $((NODE_COUNT-1))`; do
> NODE_INFO=$($PCP_NODE_INFO 1 $OCF_RESKEY_pcp_admin_host $OCF_RESKEY_pcp_admin_port $OCF_RESKEY_pcp_admin_username $OCF_RESKEY_pcp_admin_password $node | awk '{print $3}')
> if [ "${NODE_INFO}" == "3" ]; then
This is also bash-y, better use just '=':
if [ "$NODE_INFO" = "3" ]; then
> ocf_log info "Node $node is currently detached. Attempting to reattach the node."
> $PCP_ATTACH_NODE 1 $OCF_RESKEY_pcp_admin_host $OCF_RESKEY_pcp_admin_port $OCF_RESKEY_pcp_admin_username $OCF_RESKEY_pcp_admin_password $node
> ATTACHED="1"
> fi
> done
> if [ -n "${ATTACHED}" ]; then
> sleep 1
> fi
> fi
> # Fail if configured to fail on one or more detached nodes and a node is still detached
> if [ "${OCF_RESKEY_fail_on_detached}" == "true" ] ||
> [ "${OCF_RESKEY_fail_on_detached}" == "t" ] ||
> [ "${OCF_RESKEY_fail_on_detached}" == "yes" ] ||
> [ "${OCF_RESKEY_fail_on_detached}" == "y" ]; then
> NODE_COUNT=$($PCP_NODE_COUNT 1 $OCF_RESKEY_pcp_admin_host $OCF_RESKEY_pcp_admin_port $OCF_RESKEY_pcp_admin_username $OCF_RESKEY_pcp_admin_password)
> for (( node=0; $node < $NODE_COUNT; node=(($node+1)) )); do
> NODE_INFO=$($PCP_NODE_INFO 1 $OCF_RESKEY_pcp_admin_host $OCF_RESKEY_pcp_admin_port $OCF_RESKEY_pcp_admin_username $OCF_RESKEY_pcp_admin_password $node | awk '{print $3}')
> if [ "${NODE_INFO}" == "3" ]; then
> ocf_log error "Node $node is detached. The pgpool-II service has failed."
> return $OCF_ERR_GENERIC
> fi
> done
> fi
> # Fail if one of the specifically configured nodes is detached at this point
> if [ -n "${OCF_RESKEY_fail_on_node_detached}" ]; then
> NODE_COUNT=$($PCP_NODE_COUNT 1 $OCF_RESKEY_pcp_admin_host $OCF_RESKEY_pcp_admin_port $OCF_RESKEY_pcp_admin_username $OCF_RESKEY_pcp_admin_password)
> for (( node=0; $node < $NODE_COUNT; node=(($node+1)) )); do
> NODE_INFO=$($PCP_NODE_INFO 1 $OCF_RESKEY_pcp_admin_host $OCF_RESKEY_pcp_admin_port $OCF_RESKEY_pcp_admin_username $OCF_RESKEY_pcp_admin_password $node | awk '{print $3}')
> if [ "${NODE_INFO}" == "3" ]; then
> TOKEN=${OCF_RESKEY_fail_on_node_detached%%,*}
> TOKEN_STRING=${OCF_RESKEY_fail_on_node_detached#*,}
> while [ -n $TOKEN ] && [ "${TOKEN}" != "${TOKEN_STRING}" ]; do
You have to protect $TOKEN:
while [ -n "$TOKEN" ] && [ "${TOKEN}" != "${TOKEN_STRING}" ]; do
And please use {} sparingly, they just make the code harder to
read.
> if [ $TOKEN -eq $node ]; then
Here too.
> ocf_log error "Node $node is detached. The pgpool-II service has failed."
> return $OCF_ERR_GENERIC
> fi
> TOKEN=${TOKEN_STRING%%,*}
> if [ "${TOKEN_STRING}" == "${TOKEN_STRING#*,}" ]; then
> TOKEN_STRING=""
> else
> TOKEN_STRING=${TOKEN_STRING#*,}
> fi
> done
The bash substition/parameter expansion is too cryptic. To me at
least. Please use sed if possible.
> fi
> done
> fi
> fi
> # Service is running and there is no reason to fail
> return $OCF_SUCCESS
> }
Please rewrite pgpool2_status. There is a lot of repetition on
both the line level (NODE_COUNT, etc) and the if constructs are
equal for the most part. Try to stuff common strings into
variables.
> pgpool2_validate() {
> # If we're running as a clone, are the clone meta attrs OK?
> # if [ "${OCF_RESKEY_CRM_meta_clone}" ]; then
> # if [ "${OCF_RESKEY_CRM_meta_clone_node_max}" != 1 ]; then
> # ocf_log error "Misconfigured clone parameters. Must set meta attribute \"clone_node_max\" to 1, got ${OCF_RESKEY_CRM_meta_clone_node_max}."
> # return $OCF_ERR_ARGS
> # fi
> # fi
> if [ -z "${OCF_RESKEY_pcp_admin_username}" ]; then
> ocf_log error "Missing required parameter \"pcp_admin_username\"."
> return $OCF_ERR_ARGS
> fi
> if [ -z "${OCF_RESKEY_pcp_admin_password}" ]; then
> ocf_log error "Missing required parameter \"pcp_admin_password\"."
> return $OCF_ERR_ARGS
> fi
> if [ -z "${OCF_RESKEY_pcp_admin_host}" ]; then
> ocf_log error "Missing required parameter \"pcp_admin_host\"."
> return $OCF_ERR_ARGS
> fi
pcp_admin_host has a default as well as the following parameters.
> if [ -z "${OCF_RESKEY_pcp_admin_port}" ]; then
> ocf_log error "Missing required parameter \"pcp_admin_port\"."
> return $OCF_ERR_ARGS
> fi
> # Did we get a path for the pgpool binary?
> if [ -z "${OCF_RESKEY_pgpool_bin}" ]; then
> ocf_log error "Missing required parameter \"pgpool_bin\"."
> return $OCF_ERR_ARGS
> else
> if [ -x "$PGPOOL" ]; then
> ocf_log error "The pgpool binary is not executable or is not installed."
> return $OCF_ERR_INSTALLED
> fi
But it's good to check if the binary exists.
> fi
> # Did we get a path for the pcp_attach_node binary?
> if [ -z "${OCF_RESKEY_pcp_attach_node_bin}" ]; then
> ocf_log error "Missing required parameter \"pcp_attach_node_bin\"."
> return $OCF_ERR_ARGS
> else
> if [ -x "$PCP_ATTACH_NODE" ]; then
> ocf_log error "The pcp_attach_node binary is not executable or is not installed."
> return $OCF_ERR_INSTALLED
> fi
> fi
> # Did we get a path for the pcp_detach_node binary?
> if [ -z "${OCF_RESKEY_pcp_detach_node_bin}" ]; then
> ocf_log error "Missing required parameter \"pcp_detach_node_bin\"."
> return $OCF_ERR_ARGS
> else
> if [ -x "$PCP_DETACH_NODE" ]; then
> ocf_log error "The pcp_detach_node binary is not executable or is not installed."
> return $OCF_ERR_INSTALLED
> fi
> fi
> # Did we get a path for the pcp_node_count binary?
> if [ -z "${OCF_RESKEY_pcp_node_count_bin}" ]; then
> ocf_log error "Missing required parameter \"pcp_node_count_bin\"."
> return $OCF_ERR_ARGS
> else
> if [ -x "$PCP_NODE_COUNT" ]; then
> ocf_log error "The pcp_node_count binary is not executable or is not installed."
> return $OCF_ERR_INSTALLED
> fi
> fi
> # Did we get a path for the pcp_node_info binary?
> if [ -z "${OCF_RESKEY_pcp_node_info_bin}" ]; then
> ocf_log error "Missing required parameter \"pcp_node_info_bin\"."
> return $OCF_ERR_ARGS
> else
> if [ -x "$PCP_NODE_INFO" ]; then
> ocf_log error "The pcp_node_info binary is not executable or is not installed."
> return $OCF_ERR_INSTALLED
> fi
> fi
> if [ -n "${OCF_RESKEY_stop_mode}" ]; then
> if [ "${OCF_RESKEY_stop_mode}" != "f" ] &&
> [ "${OCF_RESKEY_stop_mode}" != "s" ] &&
> [ "${OCF_RESKEY_stop_mode}" != "i" ] &&
> [ "${OCF_RESKEY_stop_mode}" != "fast" ] &&
> [ "${OCF_RESKEY_stop_mode}" != "smart" ] &&
> [ "${OCF_RESKEY_stop_mode}" != "immediate" ]; then
> ocf_log error "Stop mode is invalid."
> return $OCF_ERR_ARGS
> fi
Perhaps use grep -E in these situations:
if echo "$OCF_RESKEY_stop_mode" | grep -E '[fsi]|fast|start|immediate'; then
or case:
case "$OCF_RESKEY_stop_mode" in
f|s|i|fast|start|immediate) break;;
*) error....
esac
> else
> ocf_log error "Stop mode was not specified."
> return $OCF_ERR_ARGS
> fi
> if [ -n "${OCF_RESKEY_auto_reconnect}" ] &&
> [ "${OCF_RESKEY_auto_reconnect}" != "true" ] &&
> [ "${OCF_RESKEY_auto_reconnect}" != "false" ]; then
Better with is_ocf_true
> ocf_log error "Parameter 'auto_reconnect' must be empty, 'true', or 'false'."
> return $OCF_ERR_ARGS
> fi
> if [ -n "${OCF_RESKEY_fail_on_detached}" ] &&
> [ "${OCF_RESKEY_fail_on_detached}" != "true" ] &&
> [ "${OCF_RESKEY_fail_on_detached}" != "false" ]; then
> ocf_log error "Parameter 'fail_on_detached' must be empty, 'true', or 'false'."
> return $OCF_ERR_ARGS
> fi
> shopt -s extglob
> if [ -n "${OCF_RESKEY_fail_on_node_detached}" ]; then
> TOKEN=${OCF_RESKEY_fail_on_node_detached%%,*}
> TOKEN_STRING=${OCF_RESKEY_fail_on_node_detached#*,}
> while [ -n $TOKEN ] && [ "${TOKEN}" != "${TOKEN_STRING}" ]; do
> case $TOKEN in
> [^0-9])
> ocf_log error "Invalid token '${TOKEN}' in parameter 'fail_on_node_detached'."
> return $OCF_ERR_ARGS
> ;;
> esac
> TOKEN=${TOKEN_STRING%%,*}
> if [ "${TOKEN_STRING}" == "${TOKEN_STRING#*,}" ]; then
> TOKEN_STRING=""
> else
> TOKEN_STRING=${TOKEN_STRING#*,}
> fi
> done
> fi
Oh, again extglob and friends. Please try with sed.
> return $OCF_SUCCESS
> }
>
> # These two actions must always succeed
> case $__OCF_ACTION in
> meta-data) meta_data
> # OCF variables are not set when querying meta-data
> exit 0
> ;;
> usage|help) pgpool2_usage
> exit $OCF_SUCCESS
> ;;
> esac
>
> pgpool2_validate || exit $?
>
> PGPOOL=$OCF_RESKEY_pgpool_bin
> PCP_ATTACH_NODE=$OCF_RESKEY_pcp_attach_node_bin
> PCP_DETACH_NODE=$OCF_RESKEY_pcp_detach_node_bin
> PCP_NODE_COUNT=$OCF_RESKEY_pcp_node_count_bin
> PCP_NODE_INFO=$OCF_RESKEY_pcp_node_info_bin
> STOP_MODE=$OCF_RESKEY_stop_mode
>
> case $__OCF_ACTION in
> start) pgpool2_start;;
> stop) pgpool2_stop;;
> status|monitor) pgpool2_status;;
> reload) ocf_log info "Reloading..."
> pgpool2_start
"reload" should offer something more than starting a stopped
instance. If it can't, then just drop the action from the agent,
it's not required.
Many thanks for the contribution!
Cheers,
Dejan
> ;;
> validate-all) ;;
> *) pgpool2_usage
> exit $OCF_ERR_UNIMPLEMENTED
> ;;
> esac
> rc=$?
> ocf_log debug "${OCF_RESOURCE_INSTANCE} $__OCF_ACTION returned $rc"
> exit $rc
>
>
>
> Eliot Gable
> Senior Product Developer
> 1228 Euclid Ave, Suite 390
> Cleveland, OH 44115
>
> Direct: 216-373-4808
> Fax: 216-373-4657
> egable [at] broadvox
>
>
> CONFIDENTIAL COMMUNICATION. This e-mail and any files transmitted with it are confidential and are intended solely for the use of the individual or entity to whom it is addressed. If you are not the intended recipient, please call me immediately. BROADVOX is a registered trademark of Broadvox, LLC.
>
>
>
> CONFIDENTIAL. This e-mail and any attached files are confidential and should be destroyed and/or returned if you are not the intended and proper recipient.
>
> _______________________________________________
> Pacemaker mailing list: Pacemaker [at] oss
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
_______________________________________________________
Linux-HA-Dev: Linux-HA-Dev [at] lists
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
|
