pmReconnectContext — reconnect to a PMAPI context

C Synopsis

#include <pcp/pmapi.h>

int pmReconnectContext(int handle);

cc ... -lpcp

Description

As a consequence of network, host or Performance Metrics Collector Daemon (PMCD) failures, an application's connection to a PMCD may be established and then subsequently lost.

The routine pmReconnectContext allows an application to request that the context identified by handle should be re-established, provided the associated metrics source is accessible.

When the source of metrics associated with the context handle is pmcd(1), then to avoid flooding the system with reconnect requests, pmReconnectContext will only attempt a reconnection after a suitable delay from the previous unsuccessful attempt to reconnect this context. This imposed restriction on the reconnect re-try time interval uses an exponential back-off so that the  initial delay is 5 seconds after the first unsuccessful attempt, then 10 seconds, then 20 seconds, then 40 seconds and then 80 seconds thereafter.

The environment variable PMCD_RECONNECT_TIMEOUT may be used to redefine the back-off intervals, see PMAPI(3).

Calling pmReconnectContext with a handle identifying a currently connected pmcd(1) context will cause the connection to be broken before any reconnection is attempted.

If handle identifies a context associated with an archive source of metrics, pmReconnectContext returns without delay.

If the reconnection succeeds, pmReconnectContext returns handle.

As a side-effect of reconnecting, any derived metrics that have previously been defined using pmRegisterDerived(3), pmRegisterDerivedMetric(3) or pmLoadDerivedConfig(3) will be re-processed and re-bound to the available metrics from the reconnected source. The support of dynamic definition for derived metrics provides one use case where pmReconnectContext may be called even if the connection to the metrics source has not been lost.

Note that even in the case of a successful reconnection, pmReconnectContext does not change the current Performance Metrics Application Programming Interface (PMAPI) context, so handle remains valid.

When attempting to connect to a remote pmcd(1) on a machine that is booting, pmReconnectContext could potentially block for a long time until the remote machine finishes its initialization. pmReconnectContext will abort and return an error if the connection has not been established after some specified interval has elapsed.  The default interval is 5 seconds.  This may be modified by setting PMCD_CONNECT_TIMEOUT in the environment to a real number of seconds for the desired timeout. This is most useful in cases where the remote host is at the end of a slow network, requiring longer latencies to establish the connection correctly.

Environment

PMCD_CONNECT_TIMEOUT

Timeout period (in seconds) for pmcd(1) connection attempts.

PMCD_RECONNECT_TIMEOUT

Redefines the back-off intervals - refer to PMAPI(3).

See Also

pmcd(1), PMAPI(3), pmLoadDerivedConfig(3), pmNewContext(3), pmRegisterDerived(3), pmRegisterDerivedMetric(3) and pmUseContext(3).

Diagnostics

PM_ERR_NOCONTEXT

handle does not identify a valid PMAPI context

-ETIMEDOUT

The re-try time has not elapsed, or the reconnection is attempted and fails.

Caveat

Applications that use gethostbyname(3) should exercise caution because the static fields in struct hostent may not be preserved across some PMAPI(3) calls. In particular, pmNewContext(3) and pmReconnectContext(3) both may call gethostbyname(3) internally.

Referenced By

pmNewContext(3), pmRegisterDerived(3).

The man page pmreconnectcontext(3) is an alias of pmReconnectContext(3).

PCP Performance Co-Pilot