Operating system name and version: Red Hat Enterprise Linux 5.4 64bit
Version of smsd: 3-3.1.6
Smsd installed from: sources
Name and model of a modem / phone: Wavecom Fastrack M1306B
Interface: serial

Hi,

Im running OpenNMS with sms notification (triggering sendsms on alert). Everything works well for the past month until more interfaces being added for monitoring in OpenNMS. Problem is when there's alot of alert coming in from OpenNMS and sendsms script spawning processes and then turn to sleep mode. At that moment smsd is still running but the logs are silent as if the process is in hang state.

From top command
top - 10:41:24 up 23 days, 21:59, 3 users, load average: 0.29, 0.49, 0.45 Tasks: 3646 total, 1 running, 3644 sleeping, 1 stopped, 0 zombie Cpu(s): 1.5%us, 0.4%sy, 0.0%ni, 97.7%id, 0.3%wa, 0.0%hi, 0.1%si, 0.0% Mem: 32950808k total, 21663364k used, 11287444k free, 1260964k buffers Swap: 16386292k total, 0k used, 16386292k free, 16855132k cached

From ps command
root 30330 0.0 0.0 65980 1204 pts/3 S 10:53 0:00 /bin/sh /usr root 30335 0.0 0.0 65980 1204 pts/3 S 10:53 0:00 /bin/sh /usr root 30338 0.0 0.0 65980 1204 pts/3 S 10:53 0:00 /bin/sh /usr root 30340 0.0 0.0 65980 1204 pts/3 S 10:53 0:00 /bin/sh /usr root 30344 0.0 0.0 65980 1204 pts/3 S 10:53 0:00 /bin/sh /usr root 30347 0.0 0.0 65980 1200 pts/3 S 10:53 0:00 /bin/sh /usr

After when notification function in OpenNMS are turned off, the spawned script slowly died and I restarted smsd then everything works.

Config file being used.
# Global configuration devices = WAVECOM logfile = /var/log/smstools/smsd.log infofile = /var/run/smstools/smsd.working pidfile = /var/run/smstools/smsd.pid outgoing = /var/spool/sms/outgoing checked = /var/spool/sms/checked failed = /var/spool/sms/failed incoming = /var/spool/sms/incoming sent = /var/spool/sms/sent stats = /var/log/smstools/smsd_stats loglevel = critical #blacklist = /etc/smstools/blacklist #whitelist = /etc/smstools/whitelist autosplit = 3 # Modem configuration [WAVECOM] device = /dev/ttyS0 incoming = yes baudrate = 115200 report = yes mode = new smsc = 60193900000

The question is, are there any queue function or delay in sending sms from sendsms script or from the config file to handle a storm of alerts. Thanks.

When the mainprocess of smsd was stopped, what was told in the log file?

There are no many reasons why the mainprocess will stop. The most likely reason is that the process cannot move the file. As all permissions are checked during the startup, moving should work without problems. Since the problem is related to the high number of messages, would it be possible that the outgoing message file name is always the same? In this case the smsd could detect that the moving has failed because a file with the same name still exists (because it's re-created).

You could use loglevel = notice when resolving this issue.

insolutions wrote

The question is, are there any queue function or delay in sending sms from sendsms script or from the config file to handle a storm of alerts.

No there is not. But usually it's not required, as all possible processes can create message files as much as they want and smsd just handles them. Large number of files does not cause problems.

Are you using the original sendsms script, or is it modified somehow?

Im using an unmodified version of the script. I will try to set loglevel to notice and see how it goes.

In the other hand the logs seem okay with no visible error, its just it has been stucked for 2 days with no activity and thats how i assume the main process is in hang state. I will try to acquire the tail logs of it, when it is still in hang state and after I restarted the mainprocess.

And as for the permission, Im running it as root and would it have any permission problems, since the initialization in the logs doesn't spit out any error messages.