starlingx/metal
Code Issues Proposed changes

Improve BMC password first fetch handling in hwmon

Browse Source 
Trying to get the BMC password through barbican before
the ping succeeds leads to an early bmc access lost
failure that
 1. produces a misleading bmc access lost failure log ;
    bmc access had not even been established yet.
 2. imposes as retry wait that delays re-establishing
    bmc access and therefore overall sensor monitoring.

This update also

  1. adds hostname to some of the secretUtil  API
     interfaces so that logs ar reported against the
     correct host rather than always the current
     controller hostname.

   2. Changes some success path logging to dlogs to
      reduce log noise.

   3. simplifies a ping ok log

Change-Id: Ib3b7de212294d6dc350ee17d363f4009b3b0dcb0
Story: 2005861
Task: 36595
Signed-off-by: Eric MacDonald <eric.macdonald@windriver.com>
This commit is contained in:
Eric MacDonald 2019-09-12 09:59:13 -04:00
parent 60f227c948
commit 0d63a16d8d
5 changed files with 47 additions and 39 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
mtce-common/src/common/pingUtil.cpp
View File

@ -504,14 +504,11 @@ int pingUtil_recv ( ping_info_type & ping_info,
* are in monitoring mode. */
if ( ping_info.monitoring == false )
{
/* ... only want the log when we ar first connecting */
ilog ("%s ping recv %s ok ; identity:%04x sequence:%04x (try %d) (%d)\n",
/* ... only want the log when we are first connecting */
ilog ("%s ping to %s ok ; (try %d)",
ping_info.hostname.c_str(),
ping_info.ip.c_str(),
ping_info.identity,
ping_info.sequence,
ping_info.recv_retries+1,
i);
ping_info.recv_retries+1);
}
else
{

29
mtce-common/src/common/secretUtil.cpp
View File

@ -52,6 +52,7 @@ barbicanSecret_type * secretUtil_find_secret ( string & host_uuid )
}
barbicanSecret_type * secretUtil_manage_secret ( libEvent & event,
string & hostname,
string & host_uuid,
struct mtc_timer & secret_timer,
void (*handler)(int, siginfo_t*, void*))
@ -77,7 +78,7 @@ barbicanSecret_type * secretUtil_manage_secret ( libEvent & event,
{
if ( secret_timer.ring == true )
{
rc = secretUtil_get_secret ( event, host_uuid );
rc = secretUtil_get_secret ( event, hostname, host_uuid );
if (rc)
{
wlog ( "%s getting secret reference failed \n", host_uuid.c_str() );
@ -100,7 +101,7 @@ barbicanSecret_type * secretUtil_manage_secret ( libEvent & event,
{
if ( secret_timer.ring == true )
{
rc = secretUtil_read_secret ( event, host_uuid );
rc = secretUtil_read_secret ( event, hostname, host_uuid );
if (rc)
{
wlog ( "%s getting secret payload failed \n", host_uuid.c_str() );
@ -169,7 +170,9 @@ barbicanSecret_type * secretUtil_manage_secret ( libEvent & event,
*
*/
int secretUtil_get_secret ( libEvent & event, string & host_uuid )
int secretUtil_get_secret ( libEvent & event,
string & hostname,
string & host_uuid )
{
httpUtil_event_init ( &event,
host_uuid,
@ -185,11 +188,13 @@ int secretUtil_get_secret ( libEvent & event, string & host_uuid )
}
else
{
elog ("%s failed to find secret record\n", host_uuid.c_str());
elog ("%s failed to find secret record (%s)\n",
hostname.c_str(),
host_uuid.c_str());
return FAIL;
}
event.hostname = _hn();
event.hostname = hostname ;
event.uuid = host_uuid;
event.token.url = MTC_SECRET_LABEL;
@ -220,7 +225,9 @@ int secretUtil_get_secret ( libEvent & event, string & host_uuid )
*
*********************************************************************/
int secretUtil_read_secret ( libEvent & event, string & host_uuid )
int secretUtil_read_secret ( libEvent & event,
string & hostname,
string & host_uuid )
{
httpUtil_event_init ( &event,
host_uuid,
@ -242,7 +249,7 @@ int secretUtil_read_secret ( libEvent & event, string & host_uuid )
return FAIL;
}
event.hostname = _hn();
event.hostname = hostname ;
event.uuid = host_uuid;
event.token.url = MTC_SECRET_LABEL;
@ -318,12 +325,12 @@ int secretUtil_handler ( libEvent & event )
it->second.reference = json_info.secret_ref.substr( pos+1 );
if ( it->second.reference.empty() )
{
ilog ("%s no barbican secret reference found \n", hn.c_str() );
wlog ("%s no barbican secret reference found \n", hn.c_str() );
it->second.stage = MTC_SECRET__GET_PWD_RECV;
}
else
{
ilog ("%s barbican secret reference found \n", hn.c_str() );
dlog ("%s barbican secret reference found \n", hn.c_str() );
it->second.stage = MTC_SECRET__GET_REF_RECV;
}
}
@ -332,7 +339,7 @@ int secretUtil_handler ( libEvent & event )
{
if ( event.status == HTTP_NOTFOUND )
{
ilog ("%s no barbican secret payload found \n", hn.c_str() );
wlog ("%s no barbican secret payload found \n", hn.c_str() );
}
else if ( event.status != PASS )
{
@ -341,7 +348,7 @@ int secretUtil_handler ( libEvent & event )
return ( rc ) ;
}
ilog ("%s barbican secret payload found \n", hn.c_str() );
dlog ("%s barbican secret payload found \n", hn.c_str() );
it->second.payload = event.response;
it->second.stage = MTC_SECRET__GET_PWD_RECV;
}

5
mtce-common/src/common/secretUtil.h
View File

@ -52,12 +52,13 @@ using namespace std;
barbicanSecret_type * secretUtil_find_secret ( string & host_uuid );
barbicanSecret_type * secretUtil_manage_secret ( libEvent & event,
string & hostname,
string & host_uuid,
struct mtc_timer & secret_timer,
void (*handler)(int, siginfo_t*, void*) );
int secretUtil_handler ( libEvent & event );
int secretUtil_get_secret ( libEvent & event, string & host_uuid );
int secretUtil_read_secret ( libEvent & event, string & host_uuid );
int secretUtil_get_secret ( libEvent & event, string & hostname, string & host_uuid );
int secretUtil_read_secret ( libEvent & event, string & hostname, string & host_uuid );
#endif /* __INCLUDE_MTCSECRETUTIL_H__ */

42
mtce/src/hwmon/hwmonFsm.cpp
View File

@ -98,10 +98,10 @@ void hwmonHostClass::hwmon_fsm ( void )
/* Check to see if sensor monitoring for this host is disabled.
* If it is ... */
if ( host_ptr->monitor == false )
if (( host_ptr->monitor == false ) || ( host_ptr->bm_pw.empty()))
{
/* ... make sure the thread sits in the
* idle state while disabled */
* idle state while disabled or there is no pw learned yet */
if ( thread_idle ( host_ptr->bmc_thread_ctrl ) == false )
{
if ( thread_done ( host_ptr->bmc_thread_ctrl ) == true )
@ -113,10 +113,28 @@ void hwmonHostClass::hwmon_fsm ( void )
thread_kill ( host_ptr->bmc_thread_ctrl, host_ptr->bmc_thread_info );
}
}
if ( host_ptr->bm_pw.empty( ))
{
string host_uuid = hostBase.get_uuid(host_ptr->hostname);
barbicanSecret_type * secret =
secretUtil_manage_secret( host_ptr->secretEvent,
host_ptr->hostname,
host_uuid,
host_ptr->secretTimer,
hwmonTimer_handler );
if ( secret->stage == MTC_SECRET__GET_PWD_RECV )
{
host_ptr->thread_extra_info.bm_pw = host_ptr->bm_pw = secret->payload ;
}
ilog_throttled (host_ptr->empty_secret_log_throttle, 50,
"%s waiting on bm password learn", host_ptr->hostname.c_str());
}
continue ;
}
if (( host_ptr->accessible == false ) && ( host_ptr->ping_info.ok == true ))
else if (( host_ptr->accessible == false ) && ( host_ptr->ping_info.ok == true ))
{
ilog ("%s bmc is accessible\n", host_ptr->hostname.c_str());
host_ptr->accessible = true ;
@ -153,23 +171,7 @@ void hwmonHostClass::hwmon_fsm ( void )
host_ptr->thread_extra_info.bm_pw = "" ;
}
#endif
if ( host_ptr->thread_extra_info.bm_pw.empty () )
{
string host_uuid = hostBase.get_uuid(host_ptr->hostname);
wlog_throttled ( host_ptr->empty_secret_log_throttle, 20,
"%s bm password is empty ; learning and forcing reconnect\n",
host_ptr->hostname.c_str());
barbicanSecret_type * secret = secretUtil_manage_secret( host_ptr->secretEvent,
host_uuid,
host_ptr->secretTimer,
hwmonTimer_handler );
if ( secret->stage == MTC_SECRET__GET_PWD_RECV )
{
host_ptr->ping_info.ok = false ;
host_ptr->thread_extra_info.bm_pw = host_ptr->bm_pw = secret->payload ;
}
}
else if ( host_ptr->accessible )
if ( host_ptr->accessible )
{
/* typical success path */
hwmonHostClass::bmc_sensor_monitor ( host_ptr );

1
mtce/src/maintenance/mtcNodeHdlrs.cpp
View File

@ -6222,6 +6222,7 @@ int nodeLinkClass::bmc_handler ( struct nodeLinkClass::node * node_ptr )
if ( node_ptr->thread_extra_info.bm_pw.empty() )
{
barbicanSecret_type * secret = secretUtil_manage_secret( node_ptr->secretEvent,
node_ptr->hostname,
node_ptr->uuid,
node_ptr->bm_timer,
mtcTimer_handler );