starlingx/metal
Code Issues Proposed changes

Fix up requirements for centos7 and bionic

Browse Source 
* Add dependencies for bionic:
  libevent-2.1
  libjson-c*

* Fix a couple of bugs setting /etc/hosts

Change-Id: Ice77cb9db8db367faa982e3113ed1c16065be896
Signed-off-by: Dean Troyer <dtroyer@gmail.com>
This commit is contained in:
Dean Troyer 2019-02-15 10:58:11 -06:00
parent ab3f2385e8
commit a1a98d3514
4 changed files with 32 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
devstack/files/debs/stx-metal
View File

@ -1,7 +1,10 @@
cppcheck
libevent-2.0
libevent-2.0 # dist:xenial
libevent-2.1 # dist:bionic
libevent-dev
libjson0
libjson0-dev
libjson0 # dist:xenial
libjson0-dev # dist:xenial
libjson-c3 # dist:bionic
libjson-c-dev # dist:bionic
resource-agents
facter

7
devstack/files/rpms/stx-metal Normal file
View File

@ -0,0 +1,7 @@
cppcheck
json-c
json-c-devel
libevent
libevent-devel
resource-agents
facter

4
devstack/lib/stx-metal
View File

@ -462,11 +462,11 @@ function configure_maintenance {
# TODO: enable controllerconfig to do this
# add floating ip hostname in hosts
if ! cat /etc/hosts | grep controller$ > /dev/null; then
if ! grep -q 'controller$' /etc/hosts; then
echo "192.168.204.2 controller" | sudo tee -a /etc/hosts
fi
# add controller-0
if ! cat /etc/hosts | grep [0-9]$'\s'controller-0$ > /dev/null; then
if ! grep -q 'controller-0$' /etc/hosts; then
echo "192.168.204.3 controller-0" | sudo tee -a /etc/hosts
fi

34
mtce/src/pmon/pmon.h
View File

@ -4,7 +4,7 @@
* SPDX-License-Identifier: Apache-2.0
*
*/
/*
* This implements the CGCS process Monitor ; /usr/local/bin/pmond
*
@ -24,7 +24,7 @@
* pmon_msg_init
* event_port_init
* pulse_port_init
*
*
* daemon_service_run
* wait for goenable signal
* pmon_send_event ( READY )
@ -35,7 +35,7 @@
* service_events
* pmon_send_pulse
* pmon_send_hostwd
*
*
* This daemon waits for a "goenabled" signal an then reads all the process
* configuration files in /etc/pmon.d and begins monitoring them accordingly.
* A process confguration file is expected to contain the following information
@ -44,30 +44,30 @@
* ...
*
* But who watches the watcher ? Well there is a built-in mechanism for that.
* A 'failing' or 'not running' Process Monitor Daemon (pmond) will lead to a
* degrade condition for that host.
* A 'failing' or 'not running' Process Monitor Daemon (pmond) will lead to a
* degrade condition for that host.
*
* Here is how it works ...
*
* Step 1: pmond is in inittab so that it will be respawned if it dies.
*
* Step 2: While running pmond periodically sends a pulse message to the
* the local heartbeat Client (hbsClient).
* the local heartbeat Client (hbsClient).
*
* Note: The hbsClient pulse response message has a flags field with 1
* bit dedicated to indicate the presence of the pmond on that host.
* bit dedicated to indicate the presence of the pmond on that host.
*
* Step 3: Every time the hbsClient receives a pmond pulse message it sets
* the pmond bit in the flags field of its pulse response.
* the pmond bit in the flags field of its pulse response.
*
* Note: So if the pmond dies it stops sending its pulse message that the
* pmond bit in the pulse response flags will not be set.
*
* Step 4: The heartbeat agent (hbsAgent) looks at the pulse response flags.
* For every response that does not contain a pmond flag it increments
* the pmond 'missing' counter for that host.
* the pmond 'missing' counter for that host.
*
* Step 5: Every time it sees the pmod flag it clears the counter.
* Step 5: Every time it sees the pmod flag it clears the counter.
* If that counter reaches PMOND_MISSING_THRESHOLD then that host
* is set to degraded. The degrade condition is cleared as soon
* as a single pmond flag is observed.
@ -129,7 +129,7 @@ using namespace std;
#define UNUSED(_x_) ((void) _x_)
#endif
#define AMON_MAGIC_NUM 0x12345678
#define AMON_MAGIC_NUM 0x12345678
#define CONFIG_DIR ((const char *)"/etc/pmon.d")
@ -144,7 +144,7 @@ using namespace std;
/* New PRCTL Flag
*
* Set/get notification for task state changes */
#define PR_DO_NOTIFY_TASK_STATE 17
#define PR_DO_NOTIFY_TASK_STATE 17
/* This is the data structure for requestion process death
* (and other state change) information. Sig of -1 means
@ -153,15 +153,15 @@ using namespace std;
* and will be updated with the previous values on every
* successful call.
*/
struct task_state_notify_info
struct task_state_notify_info
{
pid_t pid ;
int sig ;
unsigned int events;
};
/* The "events" bits in the struct correspond to the si_code values in the siginfo_t struct
* that would normally be sent along with a SIGCHLD to the parent process.
/* The "events" bits in the struct correspond to the si_code values in the siginfo_t struct
* that would normally be sent along with a SIGCHLD to the parent process.
* The bits are mapped like this:
*
* 1 << (info->si_code & 0xFF)
@ -170,7 +170,7 @@ The possible si_code values are defined in /usr/include/bits/siginfo.h and are:
enum
{
CLD_EXITED = 1, Child has exited.
CLD_EXITED = 1, Child has exited.
CLD_KILLED, Child was killed.
CLD_DUMPED, Child terminated abnormally.
CLD_TRAPPED, Traced child has trapped.
@ -239,7 +239,7 @@ void pmon_set_ctrl_ptr ( pmon_ctrl_type * ctrl_ptr );
#define PMON_RT_SIGNAL (SIGRTMIN+1)
#define PMON_EVENT_FLAGS ( MON_EXITED | MON_KILLED )
// MON_DUMPED
// MON_STOPPED
// MON_STOPPED
// MON_TRAPPED
int setup_signal_handler ( int rt_signal_num );