[Starlingx-discuss] pmond and systemd handling for fm-rest-api

[Penney, Don]

You should be able to use $MAINPID for the PID, rather than pgrep. We have a number of cases in StarlingX repos doing something like:

ExecStartPost=/bin/bash -c 'echo $MAINPID > /var/run/XXXXX.pid'

Note that those cases do not set PIDFile.

The following thread has a description of how systemd determines $MAINPID. If PIDFile is set, it would read the value from there, which would conflict with using it to write the pidfile:
https://lists.freedesktop.org/archives/systemd-devel/2013-February/009201.html


-----Original Message-----
From: Dominig ar Foll (Intel Open Source) [mailto:dominig.arfoll at fridu.net] 
Sent: Tuesday, September 10, 2019 2:35 PM
To: starlingx-discuss at lists.starlingx.io
Subject: Re: [Starlingx-discuss] pmond and systemd handling for fm-rest-api

Erich,

systemD will not create the PID file for you, the PIDFile statement is
used by systemD to locate the PID file, created by the launched service,
in order to check that the service has started and is running if the
monitor option is requested.
PIDFile is used in the case of a "type=forked".

What ever "type" is used,  the PID file is not created by systemD but by
the launched service itself.
So your issue, as you start fm-api directly, you need to get the PID
written in the proper place.

An ExecStartPost statement would likely help you.
You could try to add the following line to your service file. It should
create your missing PID file that is required by pmond.

ExecStartPost=/bin/sh -c 'umask 022; pgrep fm-api > /var/run/fm-api.pid'

Regards
Dominig
 

On 10/09/2019 17:43, Cordoba Malibran, Erich wrote:
> Hi,
>
> I'm converting the fm-api init script to be handled completely by
> systemd, however I'm seeing that, once I created the unit file, pmod is
> constantly launching fm-api (which causes a failure in fm-api is there
> is a conflict in the network address usage).
>
> So, this is my unit file for fm-api.
>
> [Unit]
> Description=Fault Management REST API Service
> After=nfscommon.service sw-patch.service
> After=network-online.target systemd-udev-settle.service
>
> [Service]
> Type=simple
> RemainAfterExit=yes
> User=root
> ExecStart=/usr/bin/fm-api --config-file=/etc/fm/fm.conf
> PIDFile=/var/run/fm-api.pid
>
> [Install]
> WantedBy=multi-user.target
>
>
> Then I edit the /etc/pmon.d/fm-api.conf file to this:
>
> [process]
> process  = fm-api
> pidfile  = /var/run/fm-api.pid
> service  = fm-api.service
> style    = systemd        ; ocf or lsb
> severity = major          ; minor, major, critical
> restarts = 3              ; restarts before error assertion
> interval = 5              ; number of seconds to wait between restarts
> debounce = 20             ; number of seconds to wait before degrade
> clear
>
>
> What I've seen is that pmond is looking for the pidfile, but this is
> not created as this was done by the deleted init script[0].
>
> Also, according to this comment[1], looking for the pidfile is the last
> resource but I can see frequent attempts to find this file, see below: 
>
> $ sudo strace -e trace=file -p 86386 
>
> --- SIGRT_2 {si_signo=SIGRT_2, si_code=SI_TIMER, si_timerid=0x20, si_overrun=0, si_value={int=6882944, ptr=0x690680}} ---                                                                     
> open("/var/run/fm-api.pid", O_RDONLY)   = -1 ENOENT (No such file or directory)                                                                                                               
> --- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=3222970, si_uid=0, si_status=0, si_utime=0, si_stime=0} ---
> --- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=3222617, si_uid=0, si_status=0, si_utime=0, si_stime=0} ---
> --- SIGRT_2 {si_signo=SIGRT_2, si_code=SI_TIMER, si_timerid=0x20, si_overrun=0, si_value={int=6882944, ptr=0x690680}} ---
> --- SIGRT_2 {si_signo=SIGRT_2, si_code=SI_TIMER, si_timerid=0x20, si_overrun=0, si_value={int=6882944, ptr=0x690680}} ---
> --- SIGRT_2 {si_signo=SIGRT_2, si_code=SI_TIMER, si_timerid=0x7efa, si_overrun=0, si_value={int=6853136, ptr=0x689210}} ---
> --- SIGRT_2 {si_signo=SIGRT_2, si_code=SI_TIMER, si_timerid=0x20, si_overrun=0, si_value={int=6882944, ptr=0x690680}} ---
> open("/var/run/fm-api.pid", O_RDONLY)   = -1 ENOENT (No such file or directory)
> --- SIGRT_2 {si_signo=SIGRT_2, si_code=SI_TIMER, si_timerid=0x7ed1, si_overrun=0, si_value={int=6883584, ptr=0x690900}} ---
>
>
> Does anybody know what can be causing this behavior ? 
>
> Thanks
>
>
> -Erich
>
>
> - [0] https://opendev.org/starlingx/fault/src/branch/master/fm-rest-api/fm/scripts/fm-api#L29
> - [1] https://opendev.org/starlingx/metal/src/branch/master/mtce/src/pmon/pmonHdlr.cpp#L1621
>
> _______________________________________________
> Starlingx-discuss mailing list
> Starlingx-discuss at lists.starlingx.io
> http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss

-- 
Dominig ar Foll
Senior Software Architect
Intel Open Source Technology Centre


_______________________________________________
Starlingx-discuss mailing list
Starlingx-discuss at lists.starlingx.io
http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss