Sidebar
endurox:v7.5.x:manuals:tprecover.8
Name
tprecover — Transaction Processing System Recovery.
DESCRIPTION
This is special ATMI system process which is started in standard way by ndrxd(8) process application server boot. This service shall be run in single instance (min=1 and max=1). The process performs monitoring of the ndrxd executable. Two checks at intervals are performed:
- Check is PID exists.
- Check the ping response from ndrxd
If number of checks overreaches max pings setting, the ndrxd is killed by -9 signal (if it was running), and then process is respawned.tprecover will start a new copy of ndrxd, which initially is put in recovery mode. During recovery mode ndrxd requests all running servers information about their services. Once learning mode is completed ndrxd will start to do sanity works, pings and other processes to keep the system in order.
NOTE: as pings and performed at intervals and ndrxd might be very busy (not a usual case), the careful calculation of ping timeout should be made, because ndrxd is doing back pings to tprecover process. Thus if tprecover waits for ping response for too long time, the ndrxd might kill the tprecover by it self. These settings are controlled in ndrxconfig.xml(5), settings <ping_time> and <ping_max>.
OPTIONS
- [-c CHECK_PERIOD]
- This setting denotes how often to perform ndrxd availability checks. Number of seconds. The default value is 5.
- [-t PING_TOUT]
- Number of seconds to wait for ndrxd to respond to ping requests. The default value is 20.
- [-m PING_MAX]
- Max number of failed attempts of process existence tests or bad ping responses. If the number if overreached, the ndrxd gets killed and respawned.