See my second patch this morning, to get_ganglia.php. Basically added
another check for clusters existance.
-fds
On Dec 14, 2004, at 1:17 PM, Jason A. Smith wrote:
Hi Federico,
So you setup the 2.5.7 webfrontend reading from a gmetad with the
scalable option set so it doesn't output the grid tag? When I had that
setup the other day I would get this error when viewing a cluster:
<H4>Ganglia cannot find a data source. Is gmond running?</H4>
I did forget to mention that it would only error when in the cluster
view, not the grid view. I didn't submit a patch for this though, I am
not sure if it would break anything. Eliminating the code from
get_ganglia.php that checks for no grid tag and generates that error
appeared to still work for me, but I don't know if it would break
anything else. Is there a good reason for that check?
~Jason
On Tue, 2004-12-14 at 14:21, Federico Sacerdoti wrote:
Got it tested and into CVS. I see no parse errors when the scalable
mode is off.
-Federico
On Dec 14, 2004, at 8:33 AM, Jason A. Smith wrote:
Besides the TN fix, it is the only thing that has changed in gmetad's
server.c file since version 2.5.4. I attached a patch which reverts
it
back to the 2.5.4 behavior of making the scalable option determine if
gmetad will write the grid tags also. Maybe the parsing & writing of
the grid tags should be separate config options???
~Jason
On Mon, 2004-12-13 at 14:25, Federico Sacerdoti wrote:
Not sure when this happened. If you could provide a patch to get
these
behaviors in the shape you think they need to be, we will get it
into
CVS.
-Federico
On Dec 13, 2004, at 10:25 AM, Jason A. Smith wrote:
It probably isn't common, but using an old gmetad with the new
2.5.7
webfrontend scripts will cause an error from this code in
get_ganglia.php which is assuming a grid tag must be present:
# If we have no child data sources, assume something is wrong.
if (!count($grid))
{
print "<H4>Ganglia cannot find a data source. Is gmond
running?</H4>";
exit;
}
Shouldn't the error message say gmetad not gmond? Is a grid tag
really
necessary for the webfrontend to function? If I comment this code
out
the web page looks fine to me.
~Jason
On Mon, 2004-12-13 at 13:02, Jason A. Smith wrote:
I just noticed a change in gmetad's behavior with respect to the
scalable config option that happened in version 2.5.5, which I
didn't
notice before. The scalable option now only affects how gmetad
processes xml data it reads, not how it writes its own xml data.
The
documentation in the gmetad.conf file says that setting scalable
to
off
will also make it not wrap its xml output in a grid tag.
Either the documentation or code should probably be corrected.
~Jason
--
/------------------------------------------------------------------
\
| Jason A. Smith Email: [EMAIL PROTECTED]
|
| Atlas Computing Facility, Bldg. 510M Phone: (631)344-4226
|
| Brookhaven National Lab, P.O. Box 5000 Fax: (631)344-7616
|
| Upton, NY 11973-5000
|
\------------------------------------------------------------------
/
Federico
Rocks Cluster Group, San Diego Supercomputer Center, CA
--
/------------------------------------------------------------------\
| Jason A. Smith Email: [EMAIL PROTECTED] |
| Atlas Computing Facility, Bldg. 510M Phone: (631)344-4226 |
| Brookhaven National Lab, P.O. Box 5000 Fax: (631)344-7616 |
| Upton, NY 11973-5000 |
\------------------------------------------------------------------/
diff -uNr ganglia-monitor-core-2.5.6/gmetad/server.c
ganglia-monitor-core-2.5.4/gmetad/server.c
--- ganglia-monitor-core-2.5.6/gmetad/server.c 2003-09-11
16:58:46.000000000 -0400
+++ ganglia-monitor-core-2.5.4/gmetad/server.c 2003-05-22
20:01:46.000000000 -0400
@@ -203,6 +203,8 @@
rc = xml_print(client, "<GANGLIA_XML VERSION=\"%s\"
SOURCE=\"gmetad\">\n",
VERSION);
+ if (!gmetad_config.scalable_mode) return rc;
+
rc = xml_print(client, "<GRID NAME=\"%s\" AUTHORITY=\"%s\"
LOCALTIME=\"%u\">\n",
gmetad_config.gridname, getfield(root.strings,
root.authority_ptr), time(0));
@@ -213,7 +215,10 @@
int
root_report_end(client_t *client)
{
- return xml_print(client, "</GRID>\n</GANGLIA_XML>\n");
+ if (gmetad_config.scalable_mode)
+ return xml_print(client, "</GRID>\n</GANGLIA_XML>\n");
+
+ return xml_print(client, "</GANGLIA_XML>\n");
}
Federico
Rocks Cluster Group, San Diego Supercomputer Center, CA
--
/------------------------------------------------------------------\
| Jason A. Smith Email: [EMAIL PROTECTED] |
| Atlas Computing Facility, Bldg. 510M Phone: (631)344-4226 |
| Brookhaven National Lab, P.O. Box 5000 Fax: (631)344-7616 |
| Upton, NY 11973-5000 |
\------------------------------------------------------------------/
-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real
users.
Discover which products truly live up to the hype. Start reading now.
http://productguide.itmanagersjournal.com/
_______________________________________________
Ganglia-developers mailing list
Ganglia-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ganglia-developers
Federico
Rocks Cluster Group, San Diego Supercomputer Center, CA
