[BlueOnyx:25845] Re: 5209R/5210R/5211R: ISO images updated

[ernie]

Hi Michael,
just letting you know, I tried the latest ISO yesterday, and the change 
you made to move the admserv sock to it's own /run directory, now make 
the admserv run for me after a reboot. I still get an failed error in 
the boot.log, about the php-fpm, but it all comes up.

[FAILED] Failed to start The PHP FastCGI Process Manager.
See 'systemctl status php-fpm.service' for details.

When I do the 'systemctl status php-fpm.service' the result is fine and 
running, so there is still some timing related issue.

Could it be the creation of the  tmpfs partition that /run depends on?


- Ernie.





On 2022-12-14 10:42, Michael Stauber wrote:
> Hi Ernie,
> 
>> Ok, I have determined that the admserv not starting on reboot, is due 
>> a race conditions.
>> 
>> I don't know the cause, but I do have a work around by using delays in 
>> the [Service] definition sections of 3 services.
>> 
>> In httpd.service I added:
>> 
>> [Service]
>> ExecStartPre=/bin/sleep 3
>> 
>> In php-fpm.service I added:
>> 
>> [Service]
>> ExecStartPre=/bin/sleep 2
>> 
>> In admserv-php-fpm.service I added:
>> 
>> [Service]
>> ExecStartPre=/bin/sleep 3
>> 
>> 
>> Now everything starts clean.
> Yikes! THAT is indeed rare. I haven't seen that one before. Systemd is
> supposedly robust enough to not have these issues, so I wonder why
> you're having them.
> 
> Likewise: Some of these services should either get restarted
> themselves by Systemd or during the next "swatch" run of "Active
> Monitor".
> 
> Still: It is odd that you get that kind of a race condition in first
> place. Is there anything unusual about the hardware you're using? Or
> is this virtualized?