Introduce a reconnect_max_tries setting for NUT drivers

NEWS.adoc

@@ -58,10 +58,11 @@ https://github.com/networkupstools/nut/milestone/13
       attempts on timeout errors, simplifying error recovery. [PR #3414]
     * Increased default TCP response timeout to 2000 ms. [PR #3418]
 
+
  - `dummy-ups` driver updates:
-   * Added `authconf` driver parameter for repeater mode to control
-     authentication configuration discovery. It accepts `default`, `none`,
-     or a specific authconf file path. [issue #3329]
+    * Added `authconf` driver parameter for repeater mode to control
+      authentication configuration discovery. It accepts `default`, `none`,
+      or a specific authconf file path. [issue #3329]
 
  - `nhs_ser` driver updates:
     * Modernized serial communication and added validated settings for baud
@@ -171,6 +172,22 @@ https://github.com/networkupstools/nut/milestone/13
       chunks populated only partially with bytes we intended. These buffers are
       now more diligently pre-zeroed as their siblings were in other code paths,
       to avoid sending host stack garbage to a device. [#3529]
+    * Introduced a `reconnect_max_tries` setting for NUT drivers, so they can
+      be configured to exit after the specified number of reconnection attempts
+      (so that the OS service management facility can completely recycle the
+      driver program). Multiple drivers were updated to use the new facility
+      to consistently report `reconnect_trying()` states instead of what they
+      did before, including: `adelsystem_cbi`, `apc_modbus`, `apcsmart`,
+      `bcmxcp_usb`, `blazer_usb`, `generic_modbus`, `nhs_ser`, `nutdrv_qx`,
+      `powerman-pdu`, `powervar_cx_usb`, `richcomm_usb`, `riello_usb`,
+      `tripplite_usb`, `usbhid-ups` (it is possible that for now some of them
+      would log the situation twice, with old methods and the new, and that
+      some would abort the program if reconnection attempt fails). [PR #3541]
+    * `belkin`, `bestfortress`, `genericups`, `powerpanel`, `tripplite` drivers
+      introduced attempts to reconnect in case of failure, instead of just
+      exiting right away or remaining in data stale mode indefinitely, also
+      using `reconnect_trying()` for consistent reporting. They now track
+      serial port file descriptor validity a bit more diligently. [PR #3541]
 
  - NUT client libraries:
     * Complete support for actions documented in `docs/net-protocol.txt`

UPGRADING.adoc

@@ -74,6 +74,11 @@ The new `-A filename` option defaults to trying to use a `nutauth.conf` file
   reading one ('none' as the legacy default). See the updated manual pages
   for more details. [issues #3329, #3411]
 
+- The `powervar_cx_usb`, `tripplite_usb` drivers used a built-in limit on
+  reconnection attempts after which they exited ('60' and '10' respectively).
+  This was revised to follow the new common setting `reconnect_max_tries`,
+  which defaults to trying indefinitely now. [#3541]
+
 Changes from 2.8.4 to 2.8.5
 ---------------------------
 

conf/ups.conf.sample

@@ -112,6 +112,25 @@ maxretry = 3
 #
 #                The default is 75 seconds.
 #
+# reconnect_max_tries: OPTIONAL.  This can be set as a global variable above
+#              your first UPS definition and it can also be set in a
+#              UPS section. If a driver loses connection to the device,
+#              try reconnecting for the specified number of iterations and
+#              exit (so some OS service management facility or an alerted
+#              sysadmin can restart the driver program "from scratch").
+#              A zero value (unreasonable but possible) means to exit as
+#              soon as the device connection was lost.
+#              A negative value (default) means to try reconnecting
+#              indefinitely (e.g. data cable fell out, or the data path
+#              to a networked UPS was lost for a significant time).
+#
+# reconnect_report_freq: OPTIONAL.  This can be set as a global variable
+#              above your first UPS definition and it can also be set in a
+#              UPS section, and used in conjunction with reconnect_max_tries
+#              to report progress (and on-going lack of success) every N
+#              attempts, to avoid flooding the logs.  A non-positive value
+#              disables such updates completely (default).
+#
 #   debug_min: OPTIONAL.  Specify a minimum debug level for all driver daemons
 #              and for the upsdrvctl tool (when specified at global level),
 #              or for this driver daemon (when specified in a driver section),

docs/man/ups.conf.txt

@@ -100,6 +100,27 @@ its own in `upsd.conf`.
 +
 Environment variable `NUT_STATEPATH` set by caller can override this setting.
 
+*reconnect_max_tries*::
+
+Optional.  If a driver loses connection to the device, try reconnecting for
+the specified number of iterations and exit (so some OS service management
+facility or an alerted sysadmin can restart the driver program "from scratch").
++
+A zero value (unreasonable but possible) means to exit as soon as the device
+connection was lost.
++
+A negative value (default) means to try reconnecting indefinitely (e.g. data
+cable fell out, or the data path to a networked UPS was lost for a significant
+time).
+
+*reconnect_report_freq*::
+
+Optional.  This can be used in conjunction with `reconnect_max_tries` to
+report progress (and on-going lack of success) every `N` attempts, to avoid
+flooding the logs.
++
+A non-positive value disables such updates completely (default).
+
 *maxstartdelay*::
 
 Optional.  Same as the UPS field of the same name, but this is the
@@ -340,6 +361,16 @@ They would not be actively killed by `upsdrvctl` after this timeout expires.
 +
 The default is 75 seconds.
 
+*reconnect_max_tries*::
+Optional.  This can be set as a global variable above your first UPS
+definition and it can also be set in a UPS section.  See explanation
+above, in the global section.
+
+*reconnect_report_freq*::
+Optional.  This can be set as a global variable above your first UPS
+definition and it can also be set in a UPS section.  See explanation
+above, in the global section.
+
 *maxretry*::
 Optional.  This can be set as a global variable above your first UPS
 definition and it can also be set in a UPS section.  See explanation

drivers/adelsystem_cbi.c

@@ -34,7 +34,7 @@
 #endif
 
 #define DRIVER_NAME	"NUT ADELSYSTEM DC-UPS CB/CBI driver (libmodbus link type: " NUT_MODBUS_LINKTYPE_STR ")"
-#define DRIVER_VERSION	"0.08"
+#define DRIVER_VERSION	"0.09"
 
 /* variables */
 static modbus_t *mbctx = NULL;							/* modbus memory context */
@@ -1330,7 +1330,7 @@ void modbus_reconnect(void)
 	int rval;
 
 	upsdebugx(1, "modbus_reconnect, trying to reconnect to modbus server");
-	dstate_setinfo("driver.state", "reconnect.trying");
+	reconnect_trying(RECONNECT_TRYING);
 
 	/* clear current modbus context */
 	modbus_close(mbctx);
@@ -1391,5 +1391,5 @@ void modbus_reconnect(void)
 /* #elif (defined NUT_MODBUS_TIMEOUT_ARG_timeval) // some un-castable type in fields */
 #endif /* NUT_MODBUS_TIMEOUT_ARG_* */
 
-	dstate_setinfo("driver.state", "quiet");
+	reconnect_trying(RECONNECT_SUCCESS);
 }

drivers/apc_modbus.c

@@ -43,7 +43,7 @@
 #endif
 
 #define DRIVER_NAME	"NUT APC Modbus driver " DRIVER_NAME_NUT_MODBUS_HAS_USB_WITH_STR " USB support (libmodbus link type: " NUT_MODBUS_LINKTYPE_STR ")"
-#define DRIVER_VERSION	"0.20"
+#define DRIVER_VERSION	"0.21"
 
 #if defined NUT_MODBUS_HAS_USB
 
@@ -955,7 +955,7 @@ static void _apc_modbus_create_reopen_matcher(void)
 
 static int _apc_modbus_reopen(void)
 {
-	dstate_setinfo("driver.state", "reconnect.trying");
+	reconnect_trying(RECONNECT_TRYING);
 
 	if (modbus_connect(modbus_ctx) < 0) {
 		upslogx(LOG_ERR, "%s: Unable to connect Modbus: %s", __func__, modbus_strerror(errno));
@@ -975,9 +975,10 @@ static int _apc_modbus_reopen(void)
 
 	is_open = 1;
 
-	dstate_setinfo("driver.state", "reconnect.updateinfo");
+	reconnect_trying(RECONNECT_UPDATEINFO);
 	_apc_modbus_read_inventory();
-	dstate_setinfo("driver.state", "quiet");
+
+	reconnect_trying(RECONNECT_SUCCESS);
 
 	return 1;
 }

drivers/apcsmart.c

@@ -38,7 +38,7 @@
 #include "apcsmart_tabs.h"
 
 #define DRIVER_NAME	"APC Smart protocol driver"
-#define DRIVER_VERSION	"3.39"
+#define DRIVER_VERSION	"3.40"
 
 #ifdef WIN32
 # ifndef ECANCELED
@@ -477,7 +477,7 @@ static ssize_t apc_read_i(char *buf, size_t buflen, int flags, const char *fn, u
 		fatalx (EXIT_FAILURE, "Error: apc_read_i called with buflen too large");
 	}
 
-	if (INVALID_FD(upsfd))
+	if (INVALID_FD_SER(upsfd))
 		return 0;
 
 	if (flags & SER_D0) {
@@ -605,7 +605,7 @@ static ssize_t apc_write_i(unsigned char code, const char *fn, unsigned int ln)
 	ssize_t ret;
 	errno = 0;
 
-	if (INVALID_FD(upsfd))
+	if (INVALID_FD_SER(upsfd))
 		return 0;
 
 	ret = ser_send_char(upsfd, code);
@@ -2301,6 +2301,13 @@ void upsdrv_initups(void)
 	}
 
 	upsfd = extrafd = ser_open(device_path);
+
+	if (INVALID_FD_SER(upsfd)) {
+		upslogx(LOG_WARNING, "%s: failed to open %s",
+			__func__, device_path);
+		/* \todo: Deal with the failure */
+	}
+
 	apc_ser_set();
 
 	/* fill length values */
@@ -2312,14 +2319,15 @@ void upsdrv_cleanup(void)
 {
 	char temp[APC_LBUF];
 
-	if (INVALID_FD(upsfd))
+	if (INVALID_FD_SER(upsfd))
 		return;
 
 	apc_flush(0);
 	/* try to bring the UPS out of smart mode */
 	apc_write(APC_GODUMB);
 	apc_read(temp, sizeof(temp), SER_TO);
 	ser_close(upsfd, device_path);
+	upsfd = ERROR_FD_SER;
 }
 
 void upsdrv_initinfo(void)
@@ -2394,8 +2402,9 @@ void upsdrv_updateinfo(void)
 
 		/* become aggressive after a few tries */
 		if (!(last_worked % 60)) {
-			upslogx(LOG_WARNING, "Trying to reconnect to the UPS");
-			dstate_setinfo("driver.state", "reconnect.trying");
+			upsdebugx(1, "%s: Trying to reconnect to the UPS", __func__);
+
+			reconnect_trying(RECONNECT_TRYING);
 
 			upsdebugx(1, "%s: call upsdrv_cleanup", __func__);
 			/* dstate_setinfo("driver.state", "cleanup.upsdrv"); */
@@ -2412,12 +2421,18 @@ void upsdrv_updateinfo(void)
 			/* dstate_setinfo("driver.state", "init.info"); */
 			upsdrv_initinfo();
 
+			if (INVALID_FD_SER(upsfd)) {
+				upsdebugx(1, "%s: upsfd remains invalid", __func__);
+				dstate_datastale();
+				return;
+			}
+
+			reconnect_trying(RECONNECT_UPDATEINFO);
 			upsdebugx(1, "%s: call upsdrv_updateinfo", __func__);
-			dstate_setinfo("driver.state", "reconnect.updateinfo");
 			/* dstate_setinfo("driver.state", "init.updateinfo"); */
 			upsdrv_updateinfo();
 
-			dstate_setinfo("driver.state", "init.quiet");
+			reconnect_trying(RECONNECT_SUCCESS);
 		}
 
 		upsdebugx(1, "%s: nudging UPS with 'Y', iteration #%d ...",

drivers/bcmxcp_usb.c

@@ -11,7 +11,7 @@
 #include <unistd.h>
 
 #define SUBDRIVER_NAME    "USB communication subdriver"
-#define SUBDRIVER_VERSION "0.28"
+#define SUBDRIVER_VERSION "0.29"
 
 /* communication driver description structure */
 upsdrv_info_t comm_upsdrv_info = {
@@ -385,7 +385,7 @@ void upsdrv_cleanup(void)
 
 void upsdrv_reconnect(void)
 {
-	dstate_setinfo("driver.state", "reconnect.trying");
+	reconnect_trying(RECONNECT_TRYING);
 
 	upsdebugx(4, "==================================================");
 	upsdebugx(4, "= device has been disconnected, try to reconnect =");
@@ -396,7 +396,12 @@ void upsdrv_reconnect(void)
 
 	upsdrv_initups();
 
-	dstate_setinfo("driver.state", "quiet");
+	if (upsdev) {
+		reconnect_trying(RECONNECT_SUCCESS);
+		/* dstate_dataok() is called in bcmxcp.c::upsdrv_updateinfo() */
+	} else {
+		dstate_datastale();
+	}
 }
 
 /* USB functions */