Re: Increasing the timeout for shepherd services

["Ludovic Courtès" <ludo@gnu.org>]

[-- Attachment #1: Type: text/plain, Size: 1208 bytes --]

Hi Konrad,

Konrad Hinsen <konrad.hinsen@fastmail.net> skribis:

> When running Guix in a docker image, I get systematic failures for
> starting several services: syslogd, nscd, openssh. It turns out that in
> all these cases the daemons are started, but so slowly that shepherd
> declares a failure before they are fully operational.
>
> I managed to fix this for syslogd by setting #:pid-file-timeout to 30
> s. But I don't see a similar parameter for nscd and openssh, and
> moreover it seems more reasonable to increase the timeout for all
> services. Is this possible? I didn't find anything in the Shepherd
> manual.

‘read-pid-file’ in the Shepherd has a hard-coded default value of 5
seconds.  You can change it for each service, but there’s currently no
way to change the default value globally.

The patch below adds a SRFI-39 parameter in the Shepherd that would
allow you to change the default PID file delay globally.

Then, on the Guix side, we could provide a configuration option
“somewhere” (where?) to set that.  Or, you could simply add a Shepherd
service that modifies that value before any other service is started.

Thoughts?

Thanks,
Ludo’.


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: Type: text/x-patch, Size: 973 bytes --]

diff --git a/modules/shepherd/service.scm b/modules/shepherd/service.scm
index bd7e379..8e7f625 100644
--- a/modules/shepherd/service.scm
+++ b/modules/shepherd/service.scm
@@ -74,6 +74,7 @@
             make-kill-destructor
             exec-command
             fork+exec-command
+            pid-file-default-delay
             read-pid-file
             make-system-constructor
             make-system-destructor
@@ -712,7 +713,12 @@ results."
 set when starting a service."
   (environ))
 
-(define* (read-pid-file file #:key (max-delay 5)
+(define pid-file-default-delay
+  ;; The default number of seconds for which 'read-pid-file' waits for PID
+  ;; files to show up.
+  (make-parameter 5))
+
+(define* (read-pid-file file #:key (max-delay (pid-file-default-delay))
                         (validate-pid? #f))
   "Wait for MAX-DELAY seconds for FILE to show up, and read its content as a
 number.  Return #f if FILE was not created or does not contain a number;