Two GKrellM Questions

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

Two GKrellM Questions

Randall Schulz
Hi,

I recently discovered that GKrellM includes client / server capabilities
wherein a GKrellM display on one system (the client) can present the
information from another (the server).

When I learned this, I set up a monitor on my primary system to display
the information for my secondary (server) box.

Two things have come up for me since I started using it this way:

1) If the server's host is rebooted, the client never picks up a new
connection to the (restarted) server. Can it be made to do so? Right
now, I must manually restart it, which is less than ideal.

2) The complement of sensors available on the remote client is not the
same as those available on the server's machine when GKrellM is run
there locally (not in client/server mode). In particular, the GPU
(video card processor) temperature is available for display for display
in GKrellM when run on the same host as that video card, but it is not
available for display when monitoring that host remotely. Why? Can I
make the GPU temperature available for the remote client? If so, how?


Thanks.


Randall Schulz
--
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Two GKrellM Questions

Sam Clemens
Randall R Schulz wrote:

> Hi,
>
> I recently discovered that GKrellM includes client / server capabilities
> wherein a GKrellM display on one system (the client) can present the
> information from another (the server).
>
> When I learned this, I set up a monitor on my primary system to display
> the information for my secondary (server) box.
>
> Two things have come up for me since I started using it this way:
>
> 1) If the server's host is rebooted, the client never picks up a new
> connection to the (restarted) server. Can it be made to do so? Right
> now, I must manually restart it, which is less than ideal.
>
> 2) The complement of sensors available on the remote client is not the
> same as those available on the server's machine when GKrellM is run
> there locally (not in client/server mode). In particular, the GPU
> (video card processor) temperature is available for display for display
> in GKrellM when run on the same host as that video card, but it is not
> available for display when monitoring that host remotely. Why? Can I
> make the GPU temperature available for the remote client? If so, how?
>

I don't have Gkrellm installed right now, but as I recall, it
provides some scripting capability for each "sensor" displayed.
I would investigate that.

>
> Thanks.
>
>
> Randall Schulz

--
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Two GKrellM Questions

Scott Jones-6
In reply to this post by Randall Schulz
On Thursday 15 May 2008, Randall R Schulz wrote:

> 1) If the server's host is rebooted, the client never picks up a new
> connection to the (restarted) server. Can it be made to do so? Right
> now, I must manually restart it, which is less than ideal.

Look in /etc/sysconfig/gkrellmd.  The relevant items are GKRELLMD_IO_TIMEOUT
and GKRELLMD_RECONNECT_TIMEOUT.

--
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Two GKrellM Questions

Randall Schulz
On Thursday 15 May 2008 17:32, Scott Jones wrote:
> On Thursday 15 May 2008, Randall R Schulz wrote:
> > 1) If the server's host is rebooted, the client never picks up a
> > new connection to the (restarted) server. Can it be made to do so?
> > Right now, I must manually restart it, which is less than ideal.
>
> Look in /etc/sysconfig/gkrellmd.  The relevant items are
> GKRELLMD_IO_TIMEOUT and GKRELLMD_RECONNECT_TIMEOUT.

That file exists, but is empty on my system. I enabled the GKrellM
daemon via the YaST "System Services (Runlevels)" module for run levels
3 and 5.

What precisely do these configuration variables mean? Where are the
documented? If Google is to be believed, only two German pages exist
that mention these configuration entries! They do look roughly like
HTML conversions of manual pages.

Nonetheless, those English translation seem to be fairly intelligible.
I'll try to use what I can gain from them to fix my issues.


Thanks.


Randall Schulz
--
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Two GKrellM Questions

Scott Jones-6
On Thursday 15 May 2008, Randall R Schulz wrote:

> On Thursday 15 May 2008 17:32, Scott Jones wrote:
> > On Thursday 15 May 2008, Randall R Schulz wrote:
> > > 1) If the server's host is rebooted, the client never picks up a
> > > new connection to the (restarted) server. Can it be made to do so?
> > > Right now, I must manually restart it, which is less than ideal.
> >
> > Look in /etc/sysconfig/gkrellmd.  The relevant items are
> > GKRELLMD_IO_TIMEOUT and GKRELLMD_RECONNECT_TIMEOUT.
>
> That file exists, but is empty on my system. I enabled the GKrellM
> daemon via the YaST "System Services (Runlevels)" module for run levels
> 3 and 5.

Weird.  It's non-empty as provided by the package from the packman repository.

> What precisely do these configuration variables mean? Where are the
> documented? If Google is to be believed, only two German pages exist
> that mention these configuration entries! They do look roughly like
> HTML conversions of manual pages.

From the file:

## Type:        integer
## Default:     0
#
# Configure gkrellm clients to disconnect from a gkrellmd server if there is
# an io-timeout interval where the client receives no input from the server.
# Use this for conditions where gkrellmd may be ungracefully terminated such
# that socket connections are not closed.
# Minimum is 2 (less than 2 for no I/O disconnecting and is the default).
GKRELLMD_IO_TIMEOUT="0"

## Type:        integer
## Default:     0
#
# Configure gkrellm clients to attempt automatic reconnects to a gkrellmd
# server every reconnect-timeout seconds after a disconnected state is
# detected.
# Disconnected states are created by normal gkrellmd shutdowns or by an
# expiring io-timeout.
# Minimum  is 2 (less than 2 for no automatic reconnecting and is the
# default).
GKRELLMD_RECONNECT_TIMEOUT="0"


They provide the --io-timeout and --reconnect-timeout commandline options to
gkrellmd.
--
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Two GKrellM Questions

Randall Schulz
On Thursday 15 May 2008 18:24, Scott Jones wrote:

> ...
>
> > > Look in /etc/sysconfig/gkrellmd.  The relevant items are
> > > GKRELLMD_IO_TIMEOUT and GKRELLMD_RECONNECT_TIMEOUT.
> >
> > That file exists, but is empty on my system. I enabled the GKrellM
> > daemon via the YaST "System Services (Runlevels)" module for run
> > levels 3 and 5.
>
> Weird.  It's non-empty as provided by the package from the packman
> repository.

Which package? There are two primary packages on Packman (for openSUSE
10.3), as far as I can tell:

gkrellm
gkrellm-daemon

I have the former installed (on the server machine). The "Description"
field says you cannot install both because the former includes the
latter (the deamon).


> > What precisely do these configuration variables mean? Where are the
> > documented? If Google is to be believed, only two German pages
> > exist that mention these configuration entries! They do look
> > roughly like HTML conversions of manual pages.
>
> From the file:

Thanks. Is that the entire configuration file?

Generally speaking, I think it's better to attach such files to prevent
the various manglings that occur along the way in email body text.


> ...
>
> They provide the --io-timeout and --reconnect-timeout commandline
> options to gkrellmd.

I'm probably confused, but aren't these more pertinent on the client
side?


Randall Schulz
--
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Two GKrellM Questions

Scott Jones-6
On Thursday 15 May 2008, Randall R Schulz wrote:

> > Weird.  It's non-empty as provided by the package from the packman
> > repository.
>
> Which package? There are two primary packages on Packman (for openSUSE
> 10.3), as far as I can tell:
>
> gkrellm
> gkrellm-daemon
>
> I have the former installed (on the server machine). The "Description"
> field says you cannot install both because the former includes the
> latter (the deamon).

I've double-checked, and it's present in both, copied
from /var/adm/fillup-templates/sysconfig.gkrellmd as part of the postinstall
process.

> > > What precisely do these configuration variables mean? Where are the
> > > documented? If Google is to be believed, only two German pages
> > > exist that mention these configuration entries! They do look
> > > roughly like HTML conversions of manual pages.
> >
> > From the file:
>
> Thanks. Is that the entire configuration file?

No, just the relevant portions.

> Generally speaking, I think it's better to attach such files to prevent
> the various manglings that occur along the way in email body text.

Perhaps.  After almost 9 years on this list, I need to purposely remind myself
that attachments are allowed now.

> > They provide the --io-timeout and --reconnect-timeout commandline
> > options to gkrellmd.
>
> I'm probably confused, but aren't these more pertinent on the client
> side?

Nope.  The daemon tells the client how often to check for a connection
whenever the timeout values change.  I don't, however, know what the
rationale behind that decision is.
--
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Two GKrellM Questions

Randall Schulz
On Thursday 15 May 2008 20:36, Scott Jones wrote:

> On Thursday 15 May 2008, Randall R Schulz wrote:
> > > Weird.  It's non-empty as provided by the package from the
> > > packman repository.
> >
> > Which package? There are two primary packages on Packman (for
> > openSUSE 10.3), as far as I can tell:
> >
> > gkrellm
> > gkrellm-daemon
> >
> > I have the former installed (on the server machine). The
> > "Description" field says you cannot install both because the former
> > includes the latter (the deamon).
>
> I've double-checked, and it's present in both, copied
> from /var/adm/fillup-templates/sysconfig.gkrellmd as part of the
> postinstall process.

Well, I have that template. I wonder why the actual config file was
empty? Some glitch during the post-install scripts? I'll probably never
know.


> ...
> >
> > Thanks. Is that the entire configuration file?
>
> No, just the relevant portions.

Right. Now that you pointed me to the template, I can see what's
available.


> ...
>
> > I'm probably confused, but aren't these more pertinent on the
> > client side?
>
> Nope.  The daemon tells the client how often to check for a
> connection whenever the timeout values change.  I don't, however,
> know what the rationale behind that decision is.

OK. That's slightly counterintuitive, but not preposterous.

Thanks for the help.


Do you know why I'm not seeing the same complement of sensors on the
remote client as on the local client for the same monitoring target?


Randall Schulz
--
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Two GKrellM Questions

Randall Schulz
On Friday 16 May 2008 06:52, Randall R Schulz wrote:
> ...
>
> Do you know why I'm not seeing the same complement of sensors on the
> remote client as on the local client for the same monitoring target?

Could it be this?:

-==--==--==--==--==--==--==--==--==--==--==--==--==--==--==-
## Type:    yesno
## Default: no
#
# Enable mbmon support for gkrellm.
# Requires the installation of the "mbmon" package as well
# as mbmon being enabled as a daemon at startup.

GKRELLMD_MBMON="no"
-==--==--==--==--==--==--==--==--==--==--==--==--==--==--==-


Unfortunately, if I set GKRELLMD_MBMON to "yes" I get a failure during
daemon start-up:

% /etc/init.d/gkrellmd restart
Shutting down FOO                                                    
done
Starting gkrellmd Bad arg: --mbmon-port=12912

Interestingly enough, running strings on the daemon binary suggests that
it does have that argument, but only accepts it under certain
circumstances, which are not met on my system. The comment (above) says
that installation of "mbmon" package as well, but I cannot find that
package, so I guess I'll have to make do without being able to remotely
monitor my GPU temperature (assuming this is why I cannot now do so).


Anyway, setting the RECONNECT_TIMEOUT fixed the main problem of having
to manually restart the client if the server is restarted.


Thanks again.

Randall Schulz
--
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]