[Bug documentation/16645] New: Useless feedback when stap fails

[Bug documentation/16645] New: Useless feedback when stap fails

[hadess at hadess dot net]

https://sourceware.org/bugzilla/show_bug.cgi?id=16645

            Bug ID: 16645
           Summary: Useless feedback when stap fails
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: documentation
          Assignee: systemtap at sourceware dot org
          Reporter: hadess at hadess dot net

http://sourceware.org/systemtap/SystemTap_Beginners_Guide/using-systemtap.html#testing
says:
run the command stap -v -e 'probe vfs.read {printf("read performed\n");
exit()}'.
etc.

Running that command yields:
$ stap -v -e 'probe vfs.read {printf("read performed\n"); exit()}'
Using a compile server.
Unable to find a suitable compile server.
No servers online to select from.
Passes: via server ? using 187912virt/3700res/2976shr/660data kb, in
0usr/0sys/2007real ms.
Passes: via server failed.  Try again with another '-v' option.

I have no idea what that "compile server" is, I just ran this command as a
normal user that's in the stapusr group (based on the results of an earlier
failure to run):
uid=1000(hadess) gid=1000(hadess) groups=1000(hadess),10(wheel),156(stapusr)
context=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023

Shouldn't the "setting up" section of the documentation mention the
requirements for running systemtap as a normal user for user-space
intrumentation? I simply couldn't figure it out...

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug documentation/16645] Useless feedback when stap fails

[dsmith at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16645

David Smith <dsmith at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dsmith at redhat dot com

--- Comment #1 from David Smith <dsmith at redhat dot com> ---
(In reply to Bastien Nocera from comment #0)
> http://sourceware.org/systemtap/SystemTap_Beginners_Guide/using-systemtap.
> html#testing says:
> run the command stap -v -e 'probe vfs.read {printf("read performed\n");
> exit()}'.
> etc.
> 
> Running that command yields:
> $ stap -v -e 'probe vfs.read {printf("read performed\n"); exit()}'
> Using a compile server.
> Unable to find a suitable compile server.
> No servers online to select from.
> Passes: via server ? using 187912virt/3700res/2976shr/660data kb, in
> 0usr/0sys/2007real ms.
> Passes: via server failed.  Try again with another '-v' option.
> 
> I have no idea what that "compile server" is, I just ran this command as a
> normal user that's in the stapusr group (based on the results of an earlier
> failure to run):
> uid=1000(hadess) gid=1000(hadess) groups=1000(hadess),10(wheel),156(stapusr)
> context=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023
> 
> Shouldn't the "setting up" section of the documentation mention the
> requirements for running systemtap as a normal user for user-space
> intrumentation? I simply couldn't figure it out...

Thanks for reporting this. I'm not sure what is going on here, why systemtap
thinks it needs to use a compile server instead of just compiling locally.

Does this system have secure boot turned on by any chance?

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug documentation/16645] Useless feedback when stap fails

[brolley at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16645

Dave Brolley <brolley at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |brolley at redhat dot com

--- Comment #2 from Dave Brolley <brolley at redhat dot com> ---
For a user who is a member of only the stapusr group, the client realizes that
this user can only run scripts compiled by a server in --unprivileged mode and
so it automatically attempts to use a server to compile the script and adds the
--unprivileged option.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug documentation/16645] Useless feedback when stap fails

[dsmith at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16645

--- Comment #3 from David Smith <dsmith at redhat dot com> ---
As Dave said, since you are only in the 'stapusr' group and not the 'stapdev'
group, you can only run "blessed" precompiled scripts from a stap compile
server or unprivileged scripts (scripts that just probe your own user exes for
instance).

The script you tried to use requires privilege (you'd need to be root or in the
'stapdev' group), since it probes all processes.

Kernel level probes (kernel.*) probes require privilege. The 'vfs.read' probe
is a probe alias that maps back to a kernel.function probe.

There is a section in the stap manpage called 'UNPRIVILEGED USERS' that
discusses this a bit.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug documentation/16645] Useless feedback when stap fails

[lberk at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16645

Lukas Berk <lberk at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |lberk at redhat dot com

--- Comment #4 from Lukas Berk <lberk at redhat dot com> ---
This might be a useful place to have an appropriate 'error::compile_server' (or
something similar) suggestion for the user. In which, mentioning that they're
an unprivileged user, how to possibly remedy that, and pointing them to the
existing, additional documentation on the compile server.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug documentation/16645] Useless feedback when stap fails

[brolley at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16645

--- Comment #5 from Dave Brolley <brolley at redhat dot com> ---
It has long been systemtap's philosphy to run fairly silently without any -v
options. Running with -v should explain that --privilege=stapusr was
automatically selected because of your group membership and that this in turn
required the use of a compile server to compile your script.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug documentation/16645] Useless feedback when stap fails

[brolley at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16645

Dave Brolley <brolley at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|systemtap at sourceware dot org    |brolley at redhat dot com

--- Comment #6 from Dave Brolley <brolley at redhat dot com> ---
It seems reasonable that if we're going to generate errors about compile
servers (or the lack thereof) that we should at least explain why you need one
and point you at the documentation for setting one up.

-- 
You are receiving this mail because:
You are the assignee for the bug.