Read-Only FileSystem - JailShell

[higherlogic]

We're getting read-only errors with cPanel users who have the updated SHELL line in their crontab, e.g.:

2013-07-01 09:20:02 Cannot open main log file "/var/log/exim_mainlog": Read-only file system: euid=0 egid=32007

2013-07-01 09:20:02 cwd=/home/<cpuser> 3 args: /usr/sbin/sendmail -t -i

2013-07-01 09:20:02 Cannot open main log file "/var/log/exim_mainlog": Read-only file system: euid=0 egid=32007
 exim: could not open panic log - aborting: see message(s) above

As soon you as you comment out this line, it works:

SHELL="/usr/local/cpanel/bin/jailshell"

 

[cPanelMichael]

Are you experiencing problems with specific cron jobs, or are you only curious about the output? I have reviewed a couple of internal cases that imply this is the intended behavior in some cases. Feel free to open a support ticket if you want us to take a closer look:

Submit A Ticket

You can post the ticket number here so we can update this thread with the outcome.

Thank you.

 

[higherlogic]

How is this "intended" behavior?

So if a client uses the PHP mail() function in their script, and it's setup in their crontab file, they get this in their email:

Cannot open main log file "/var/log/exim_mainlog": Read-only file system: euid=0 egid=42354
cwd=/home/cpuser/public_html 3 args: /usr/sbin/sendmail -t -i
Cannot open main log file "/var/log/exim_mainlog": Read-only file system: euid=0 egid=42354

exim: could not open panic log - aborting: see message(s) above
Content-type: text/html

This line in the crontab file has broken so many scripts now.

1. If we delete this line, will cPanel just add it back?

2. I'm assuming there's no option to disable this?

If so, guess I'll just have to make a cron script that deletes this line from every crontab file...

 

[cPanelMichael]

It's likely this behavior started after updating to cPanel 11.38 due to the changes to jailed shell. You can find more information about Jailed Shell (VirtFS) on 11.38 at:

cPanel - Jailed Shell (VirtFS)

Were you able to open a ticket so we could review the specific behavior to determine if it was related to any internal cases? I did see a couple of instances where running the following command to unmount old jail mounts resolved the issue:

/scripts/clear_orphaned_virtfs_mounts

Thank you.