Oddball error from "batch"

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Oddball error from "batch"

Karl Denninger
Note -- working fine on 11.1 and 11.2, upgraded machine to 12.0-STABLE
and everything is ok that I'm aware of *except*.....

# batch
who
df
^D

Job 170 will be executed using /bin/sh

Then the time comes and... no output is emailed to me.

In the cron log file I find:

Feb 10 16:00:00 NewFS atrun[65142]: cannot open input file
E000aa018a24c3: No such file or directory

Note that scheduled cron jobs are running as expected, and the
permissions on /var/at are correct (match exactly my 11. 1 and 11.2
boxes), and in addition of looking BEFORE the job runs the named job
number IS THERE.

[\u@NewFS /var/at/jobs]# ls -al
total 13
drwxr-xr-x  2 daemon  wheel    5 Feb 10 15:55 .
drwxr-xr-x  4 root    wheel    5 Oct  8  2013 ..
-rw-r--r--  1 root    wheel    6 Feb 10 15:55 .SEQ
-rw-------  1 root    wheel    0 Jul  5  2008 .lockfile
-rwx------  1 root    wheel  615 Feb 10 15:55 E000aa018a24c3

After the error.... the file isn't there.  It was removed (as one would
expect when the job is complete.)

What the blankety-blank?!

--
Karl Denninger
[hidden email] <mailto:[hidden email]>
/The Market Ticker/
/[S/MIME encrypted email preferred]/

smime.p7s (6K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Oddball error from "batch"

Karl Denninger
On 2/10/2019 16:01, Karl Denninger wrote:

> Note -- working fine on 11.1 and 11.2, upgraded machine to 12.0-STABLE
> and everything is ok that I'm aware of *except*.....
>
> # batch
> who
> df
> ^D
>
> Job 170 will be executed using /bin/sh
>
> Then the time comes and... no output is emailed to me.
>
> In the cron log file I find:
>
> Feb 10 16:00:00 NewFS atrun[65142]: cannot open input file
> E000aa018a24c3: No such file or directory
>
> Note that scheduled cron jobs are running as expected, and the
> permissions on /var/at are correct (match exactly my 11. 1 and 11.2
> boxes), and in addition of looking BEFORE the job runs the named job
> number IS THERE.
>
> [\u@NewFS /var/at/jobs]# ls -al
> total 13
> drwxr-xr-x  2 daemon  wheel    5 Feb 10 15:55 .
> drwxr-xr-x  4 root    wheel    5 Oct  8  2013 ..
> -rw-r--r--  1 root    wheel    6 Feb 10 15:55 .SEQ
> -rw-------  1 root    wheel    0 Jul  5  2008 .lockfile
> -rwx------  1 root    wheel  615 Feb 10 15:55 E000aa018a24c3
>
> After the error.... the file isn't there.  It was removed (as one would
> expect when the job is complete.)
>
> What the blankety-blank?!
Turns out it's a nasty race in the atrun code.... I have no idea why
this hasn't bit the living daylights out of lots of people before, but
it's sure biting me!

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=235657

Includes a proposed fix... :)

--
Karl Denninger
[hidden email] <mailto:[hidden email]>
/The Market Ticker/
/[S/MIME encrypted email preferred]/

smime.p7s (6K) Download Attachment