monami-users Mailing List for MonAMI - your friendly monitoring daemon
Status: Alpha
Brought to you by:
paulmillar
Menu
▾
▴
monami-users — Help and discussion on how to use MonAMI
You can subscribe to this list here.
| 2007 |
Jan
|
Feb
|
Mar
|
Apr
|
May
(1) |
Jun
|
Jul
|
Aug
(2) |
Sep
|
Oct
|
Nov
(2) |
Dec
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2008 |
Jan
|
Feb
(8) |
Mar
(13) |
Apr
(16) |
May
(13) |
Jun
(9) |
Jul
(11) |
Aug
(3) |
Sep
(4) |
Oct
(2) |
Nov
(6) |
Dec
(6) |
| 2009 |
Jan
|
Feb
|
Mar
(6) |
Apr
|
May
(9) |
Jun
(1) |
Jul
(10) |
Aug
(6) |
Sep
|
Oct
(3) |
Nov
(7) |
Dec
|
| 2010 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
(1) |
|
From: Arnau B. <arn...@pi...> - 2010-12-13 15:19:31
|
Hi all, not sure if this list already exists or someone already subscribed. I've seen no more support or develop for a long time on this great software, but anyone has ported it to SL5 x86_64? I'm trying to, but I'm facing some problems when it loads plugins... well, if someone is still here, I'll send more info. Cheers, Arnau |
|
From: Arnau B. <arn...@pi...> - 2009-11-26 08:42:58
|
On Thu, 26 Nov 2009 09:29:48 +0100 Paul Millar wrote: > Hi Arnau, Hi Paul, [...] > I feel I'm missing something: what's wrong with this? Mmm... yep, it's confusing... ok maybe this way is clearest: # wc -l /tmp/snapshot && ls -lsah /tmp/snapshot 238 /tmp/snapshot 20K -rw-rw-rw- 1 monami monami 18K Nov 26 09:36 /tmp/snapshot # wc -l /tmp/snapshot && ls -lsah /tmp/snapshot 238 /tmp/snapshot 20K -rw-rw-rw- 1 monami monami 18K Nov 26 09:38 /tmp/snapshot # wc -l /tmp/snapshot && ls -lsah /tmp/snapshot 238 /tmp/snapshot 20K -rw-rw-rw- 1 monami monami 18K Nov 26 09:40 /tmp/snapshot > HTH, > > Paul. Cheers, Arnau |
|
From: Paul M. <p.m...@ph...> - 2009-11-26 08:31:31
|
Hi Arnau, On Wednesday 25 November 2009 17:40:36 Arnau Bria wrote: > from http://monami.sourceforge.net/userguide/ch03s05.html#filelog > > "New data is appended to the end of the file" > > but with my conf, I don't see this behaviour: > > # ls -lsah /tmp/snapshot > 20K -rw-rw-rw- 1 monami monami 20K Nov 25 15:18 /tmp/snapshot > # ls -lsah /tmp/snapshot > 24K -rw-rw-rw- 1 monami monami 22K Nov 25 15:44 /tmp/snapshot > # ls -lsah /tmp/snapshot > 28K -rw-rw-rw- 1 monami monami 25K Nov 25 16:54 /tmp/snapshot I feel I'm missing something: what's wrong with this? You could also try: tail -f /tmp/snapshot That should allow you to see the file growing in size. HTH, Paul. |
|
From: Arnau B. <arn...@pi...> - 2009-11-25 17:14:51
|
Hi all, from http://monami.sourceforge.net/userguide/ch03s05.html#filelog "New data is appended to the end of the file" but with my conf, I don't see this behaviour: # ls -lsah /tmp/snapshot 20K -rw-rw-rw- 1 monami monami 20K Nov 25 15:18 /tmp/snapshot # ls -lsah /tmp/snapshot 24K -rw-rw-rw- 1 monami monami 22K Nov 25 15:44 /tmp/snapshot # ls -lsah /tmp/snapshot 28K -rw-rw-rw- 1 monami monami 25K Nov 25 16:54 /tmp/snapshot from monaim.conf.d/maui.conf: [sample] [...] write = ganglia,snapshot [...] [snapshot] filename=/tmp/snapshot Is this default behaviour? anyone faced it before? # rpm -qa|grep monami monami-0.10-1 TIA, Arnau |
|
From: Arnau B. <arn...@pi...> - 2009-11-16 11:38:42
|
On Mon, 16 Nov 2009 09:02:38 +0100 Paul Millar wrote: > Hi Arnau, Hi Paul, > Sorry for the delay ... don't worry :-) [...] > > if anyone wants to get it too: [...] > > Really, what we need is a new release of MonAMI. I agree :-) but as I was able to use torque plugin with gid info, and some extra php code that is not available at monami's site, I wanted to share all with other potentially users... > I'm using machines very generously provided by the University of > Glasgow as a build infrastructure (Many thanks guys!). They're > currently being upgraded. Once they're back I'll try to get another > release out. great! nice to hear that! > > The only problem, and here comes my question :-) is that torque > > frame does not show this new info, so new php code is needed. I > > downloaded "external" from cvs but it's already the one I have [...] > > Sorry, I thought the new PHP code was already committed there. I'll > try to get this updated. That would be great, but there's no hurry, you could release a new version with all code needed (php included)... > Cheers, > > Paul. Cheers, Arnau |
|
From: Paul M. <p.m...@ph...> - 2009-11-16 08:03:24
|
Hi Brad, On Thursday 12 November 2009 20:05:54 Brad Anderson wrote: > [...] One of our key applications is Tomcat6. When I load > the following config into MonAMI I see the parser getting 143 lines, but > none of them get published to my ganglia channel. > > #Tomcat monitoring plugin > [tomcat] > Name = tomcat6 > Username = testuser > Password = testpass The problem here is that the tomcat plugin doesn't generate a datatree based on all available information; instead, you must also specify which elements you're interested in. Here's an alternative configuration that should provide more information: [tomcat] Name = tomcat6 Username = testuser Password = testpass ThreadPool = http-8080 Connector = 8080 For more details, please see the Tomcat section of the user manual: http://monami.sourceforge.net/userguide/ch03s04.html#tomcat Please be aware that the tomcat plugin needs a "reader" for each JMX MBean class. These readers map the flat name-space of a JMX MBean to a more manageable tree structure and add some missing information: metric type (integer, string, etc) and the corresponding units (Threads, Bytes, seconds, etc). Currently there are just two readers available: ThreadPool and Connector. HTH, Paul. |
|
From: Paul M. <p.m...@ph...> - 2009-11-16 08:01:27
|
Hi Arnau, Sorry for the delay ... On Friday 16 October 2009 12:29:48 Arnau Bria wrote: > I just added torque plugin from CVS to my monami 0.10-1 rpm install. Great! As a medium-length goal, I'm trying to stabilise the plugin API. This is desirable so the plugins can be updated independently of MonAMI-core, so allowing new versions of a plugin will interop. with an existing deployment. It's good that this worked for you, but please be aware that the API is not completely finalised (e.g., the generic on-demand monitoring support is still work-in-progress) so the same trick might not work in the future with some of the plugins. > if anyone wants to get it too: [...] Really, what we need is a new release of MonAMI. I'm using machines very generously provided by the University of Glasgow as a build infrastructure (Many thanks guys!). They're currently being upgraded. Once they're back I'll try to get another release out. > The only problem, and here comes my question :-) is that torque frame > does not show this new info, so new php code is needed. I downloaded > "external" from cvs but it's already the one I have [...] Sorry, I thought the new PHP code was already committed there. I'll try to get this updated. Cheers, Paul. |
|
From: Brad A. <bra...@ga...> - 2009-11-12 19:33:14
|
All, I am evaluating using MonAMI in our production monitoring infrastructure. One of our key applications is Tomcat6. When I load the following config into MonAMI I see the parser getting 143 lines, but none of them get published to my ganglia channel. I know my ganglia channel is working because mysql and apache metrics are publishing through MonAMI just fine. If I switch the sample write to a filelog I get a timestamp and a blank line. This leads me to think the parser needs to be updated for tomcat6. Am I correct in this assumption? If so what is the preferred way of making this change? #Tomcat monitoring plugin [tomcat] Name = tomcat6 Username = testuser Password = testpass [sample] Read = tomcat6 Write = ganglia Interval = 1m # Here is what I see when I run monamid -fvv tomcat> process_msg() [anonymous sample 1> --> 2 (std) tomcat> curl contacting http://localhost:8080/manager/jmxproxy/ tomcat> jmx parser: got 143 entries Thanks in advance, Brad Anderson |
|
From: Arnau B. <arn...@pi...> - 2009-10-16 13:04:02
|
On Fri, 16 Oct 2009 12:29:48 +0200 Arnau Bria wrote: Hi again, [...] > The only problem, and here comes my question :-) is that torque frame > does not show this new info, so new php code is needed. I downloaded > "external" from cvs but it's already the one I have, I see no > reference for that new "section" in code... > > Anyone that already run that plugin could tell me where did he get php > code? I got it from scotgrid: 4 -rw-r--r-- 1 root root 4057 Oct 16 11:58 maui-graph.php 16 -rw-r--r-- 1 root root 14780 Oct 16 12:04 mg-frame-torque.php.pre 8 -rw-r--r-- 1 root root 5642 Oct 16 13:05 multiple-graphs.css 8 -rw-r--r-- 1 root root 5917 Oct 16 13:05 torque-graph.php 24 -rw-r--r-- 1 root root 21034 Oct 16 13:05 mg-frame-torque.php 28 -rw-r--r-- 1 root root 28219 Oct 16 13:08 multiple-graphs.php If anyone is interested, feel free to ask! Cheers, Arnau |
|
From: Arnau B. <arn...@pi...> - 2009-10-16 10:56:20
|
Hi all,
not sure if this list is still alive, or if someone reads it ...
Anyway, I finally got my "jobs by unix group info" on moanmi!!!
I just added torque plugin from CVS to my monami 0.10-1 rpm install.
if anyone wants to get it too:
1.-) Download CVS code
cvs -d:pserver:ano...@mo...:/cvsroot/monami login (no passwd needed)
cvs -d:pserver:ano...@mo...:/cvsroot/monami
checkout MonAMI
2.-) Compile torque pluguin (many, many deps needed, try configure and
play a little :-) ) But, from my experience:
# history |grep yum|grep install
699 yum install flex
706 yum install libxslt-devel.i386
718 yum install docbook-xsl
842 yum install zlib-devel.i386
849 yum install docbook-xml.noarch
(I did the compile on torque server, so some torque packages already
installed)
You can do some trick in order to compile only torque, but I
did compile all code.
3.-) copy new plugin to its "rpm" location:
/usr/lib/monami/torque.so
4.-) Cahnge monami's conf:
read = torque.Users, torque.Jobs, ...
^^^^^^^^^^^^
And all data will be in ganglia!
The only problem, and here comes my question :-) is that torque frame
does not show this new info, so new php code is needed. I downloaded
"external" from cvs but it's already the one I have, I see no reference
for that new "section" in code...
Anyone that already run that plugin could tell me where did he get php
code?
Cheers,
Arnau
|
|
From: Arnau B. <arn...@pi...> - 2009-10-13 09:34:45
|
Hi all, I'm still looking for toruqe unixgroup info plugin for monami. Paul told me that it's under some CVS, but I'm not able to find it... anyone could give the cvs address, please? TIA, Arnau |
|
From: Arnau B. <arn...@pi...> - 2009-08-21 09:44:08
|
Hi Paul, I've downloaded CVS version from http://monami.cvs.sourceforge.net/viewvc/monami/MonAMI. Is that the CVS you're talking about? Cause I've compiled torque plugin (only torque plugin) and copied to /usr/lib/monami/torque.so, overwriting the old one, but if I see what data it produces (doing a snapshot) I see no info about UNIX groups. Is that CVS holding 0.11 version? looks 0.10 to me.... an other question, non related to torque module. A good way to produce datatree? appart from snapshot.... TIA, Arnau |
|
From: Arnau B. <arn...@pi...> - 2009-08-20 21:38:38
|
On Thu, 20 Aug 2009 21:41:17 +0200 Paul Millar wrote: > Hi Arnau, Hi Paul, > On Thursday 20 August 2009 17:09:00 Arnau Bria wrote: > > I'm back from holiday and I'd like to continue deploying monami in > > our site. > > Welcome back! Thank you! > > Has someone already developed something similar to what I'm looking > > for? Someone expanded torque datatree? > > You're in luck! > > Aggregating job statistics by Unix group was something added some > little while ago for Scotgrid. They, too, have torque queues based > on job max wall-clock, so needed some other way of showing per-VO > statistics. > > You can see the results on the test machine: > > http://grid03.ph.gla.ac.uk/ganglia/?r=day&sg=&c=Grid+Experimental&h=grid03.ph.gla.ac.uk > > The code isn't available as part of a released version of MonAMI > yet. However, it is in CVS so it'll be part of the next release > (v0.11). Great, two good news in one mail: 1.-) I do not have to study how to implement it. 2.-) Monami still have development on it... tomorrow morning I'll start with this. > Cheers, > Paul. Thanks for your reply Paul, Arnau |
|
From: Paul M. <p.m...@ph...> - 2009-08-20 19:41:30
|
Hi Arnau, On Thursday 20 August 2009 17:09:00 Arnau Bria wrote: > I'm back from holiday and I'd like to continue deploying monami in our > site. Welcome back! > Before going to mysql/gridftp plugins, I'd like to "add" some info to > current torque plugin. > > We do not use VO dedicated queues. They are time based. So, I'd like to > add some kind of "jobs per unix group" information (how many jobs are > running/queueud depending on unix group (atlas/cms/lhcb/whatever)). > > Has someone already developed something similar to what I'm looking > for? Someone expanded torque datatree? You're in luck! Aggregating job statistics by Unix group was something added some little while ago for Scotgrid. They, too, have torque queues based on job max wall-clock, so needed some other way of showing per-VO statistics. You can see the results on the test machine: http://grid03.ph.gla.ac.uk/ganglia/?r=day&sg=&c=Grid+Experimental&h=grid03.ph.gla.ac.uk The code isn't available as part of a released version of MonAMI yet. However, it is in CVS so it'll be part of the next release (v0.11). Cheers, Paul. |
|
From: Arnau B. <arn...@pi...> - 2009-08-20 15:09:49
|
Hi all, I'm back from holiday and I'd like to continue deploying monami in our site. Before going to mysql/gridftp plugins, I'd like to "add" some info to current torque plugin. We do not use VO dedicated queues. They are time based. So, I'd like to add some kind of "jobs per unix group" information (how many jobs are running/queueud depending on unix group (atlas/cms/lhcb/whatever)). Has someone already developed something similar to what I'm looking for? Someone expanded torque datatree? TIA, Arnau |
|
From: Paul M. <p.m...@ph...> - 2009-08-19 06:21:13
|
Hi Julian, On Tuesday 18 August 2009 06:06:48 Julian Seidenberg wrote: > [...] > However, when the tree is serialized and sent to Ganglia, I worry that > it might fail because of being too long a string. This is true, but (I believe) people so far haven't hit against this problem. Do you have a specific example that you're worried about? > I have test Ganglia's gmetric tool and found it limits strings sent to it to > 1420 characters. Has anyone else tried this? I've not tried doing this, but what you say sounds reasonable. Ganglia uses UDP packets as a transport between the different sensor daemons (the gmonds) without any frame re-assembly: the whole metric must fit in a UDP packet's payload. For "normal" situations (IPv4 and no vlan tagging) UDP packets give a maximum payload of 1472 bytes. This is the same ball-park figure as you discovered and the difference is likely accountable from additional metric metadata, the xdr encoding overhead and the metric value itself. > Does MonAMI's ganglia plug-in do anything clever to get around this > limitation? No, it doesn't do anything clever; in fact, the current code might send even truncated packets under these circumstances (oops). I'm not sure what be best here. We can try to abbreviate the metric label so it fits. There's several options for this, but none are particularly appealing: truncating the end will likely omit the most significant information, truncating the beginning makes it hard to understand which overall context the metric is in, removing text somewhere within the metric name is probably best, but how to choose where to start truncating: after the first ".", exactly in the middle, ...? Another possibility would be to simply refrain from publishing those metrics (after all, we're hitting against a limitation in Ganglia) and use some other means for getting that data in; i.e., avoid having such long metric names. One idea I've had on the back-burner is how to allow people to specify derived data; that is, metrics that reflects other metrics in the datatree. The simplest derived data would be to map some sub-tree of the datatree to a new location (either a "move" or a "copy"). In effect, one could rename metrics, so long names could be made shorter. Cheers, Paul. |
|
From: Julian S. <ju...@si...> - 2009-08-18 04:51:10
|
Hi, I've been looking at MonAMI and thinking of using it in conjunction with Ganglia to monitor some servers. I like how the monitoring data can be expressed in a tree. However, when the tree is serialized and sent to Ganglia, I worry that it might fail because of being too long a string. I have test Ganglia's gmetric tool and found it limits strings sent to it to 1420 characters. Has anyone else tried this? Does MonAMI's ganglia plug-in do anything clever to get around this limitation? Regards, Julian ------- Julian Seidenberg |
|
From: Paul M. <p.m...@ph...> - 2009-07-29 07:31:23
|
Hi Arnau, On Tuesday 28 July 2009 16:17:29 Arnau Bria wrote: > mybe we crossed mails... Yup; hey ho. > [...] I found how to solve the problem, adding a "return (0)" at the end of > multiple-grpah.php. I'm happy you've found a solution, but adding "return(0);" at the end of multiple-graphs.php shouldn't be necessary. However, it does strongly suggest where the problem might be. Could you check something? Try doing: tail -5 multiple-graphs.php | hexdump -C Cheers, Paul. |
|
From: Arnau B. <arn...@pi...> - 2009-07-28 14:26:50
|
Hi Paul, mybe we crossed mails... first, thanks for this extended reply, but I found how to solve the problem, adding a "return (0)" at the end of multiple-grpah.php. thanks, Arnau |
|
From: Paul M. <p.m...@ph...> - 2009-07-28 13:53:09
|
On Tuesday 28 July 2009 10:38:59 Arnau Bria wrote: > I saw some important diff: > > bad one: > > arnau@lx-arnau ~ $ hexdump -C torque-graph.png | head -20 > 00000000 0a 89 50 4e 47 0d 0a 1a 0a 00 00 00 0d 49 48 44 > |..PNG........IHD| 00000010 52 00 00 01 8d 00 00 00 ac 08 06 00 00 00 87 > 20 |R.............. | 00000020 f0 f6 00 00 00 4a 74 45 58 74 53 6f 66 74 > 77 61 |.....JtEXtSoftwa| 00000030 72 65 00 52 52 44 74 6f 6f 6c 2c 20 54 > 6f 62 69 |re.RRDtool, Tobi| 00000040 61 73 20 4f 65 74 69 6b 65 72 20 3c > 74 6f 62 69 |as Oetiker <tobi| > > good one (generated with rrdtool) > > arnau@lx-arnau ~ $ hexdump -C rrdoutput | head -20 > 00000000 89 50 4e 47 0d 0a 1a 0a 00 00 00 0d 49 48 44 52 > |.PNG........IHDR| 00000010 00 00 02 b9 00 00 00 e9 08 06 00 00 00 20 4f > 2f |............. O/| 00000020 6a 00 00 00 4a 74 45 58 74 53 6f 66 74 77 > 61 72 |j...JtEXtSoftwar| 00000030 65 00 52 52 44 74 6f 6f 6c 2c 20 54 6f > 62 69 61 |e.RRDtool, Tobia| 00000040 73 20 4f 65 74 69 6b 65 72 20 3c 74 > 6f 62 69 40 |s Oetiker <tobi@| Ah, this is looking promising. > notice that bad one starts with ..PNG and the good one with .PNG (0a > 89 50 vs 89 50) Indeed > [...] What could be insertering that blank line? I can't see what could be inserting that extra line before the RRD graph. However, we can now start to bisect the problem. Could you try adding the line: echo "1"; immediately before the call to passthru() in multiple-graphs.php? With this in place, try capturing the graph again and examining the contents using hexdump. Does the "1" appear before or after the new-line character (0x0a)? Assuming the "1" appears after the 0x0a character, introduce a new line (e.g., echo "2";) right at the beginning of the torque_graph.php file and repeat the process. Assuming this is before the 0x0a character, you should be able to add more "echo" lines that will progressively limit the area in which this stray 0x0a character could come from. Another thing (a quick check) could you see whether the headers are set correctly? Try getting the PNG file using wget and the "-S" option: wget -O/dev/null -S http://www.example.org/ganglia/... You'll notice that cmd_go() attempts to send some HTTP headers. This will not work if something has already emitting some output, so the presence (or otherwise) of the HTTP header: Cache-Control: no-cache, must-revalidate should indicate whether the echo/print is happening before the call to cmd_go(). Cheers, Paul. |
|
From: Arnau B. <arn...@pi...> - 2009-07-28 13:20:52
|
Hi all,
> arnau@lx-arnau ~ $ hexdump -C torque-graph.png | head -20
> 00000000 0a 89 50 4e 47 0d 0a 1a 0a 00 00 00 0d 49 48 44
> |..PNG........IHD| 00000010 52 00 00 01 8d 00 00 00 ac 08 06 00 00
> 00 87 20 |R.............. | 00000020 f0 f6 00 00 00 4a 74 45 58 74
> 53 6f 66 74 77 61 |.....JtEXtSoftwa| 00000030 72 65 00 52 52 44 74
> 6f 6f 6c 2c 20 54 6f 62 69 |re.RRDtool, Tobi| 00000040 61 73 20 4f
> 65 74 69 6b 65 72 20 3c 74 6f 62 69 |as Oetiker <tobi|
>
> good one (generated with rrdtool)
>
> arnau@lx-arnau ~ $ hexdump -C rrdoutput | head -20
> 00000000 89 50 4e 47 0d 0a 1a 0a 00 00 00 0d 49 48 44 52
> |.PNG........IHDR| 00000010 00 00 02 b9 00 00 00 e9 08 06 00 00 00
> 20 4f 2f |............. O/| 00000020 6a 00 00 00 4a 74 45 58 74 53
> 6f 66 74 77 61 72 |j...JtEXtSoftwar| 00000030 65 00 52 52 44 74 6f
> 6f 6c 2c 20 54 6f 62 69 61 |e.RRDtool, Tobia| 00000040 73 20 4f 65
> 74 69 6b 65 72 20 3c 74 6f 62 69 40 |s Oetiker <tobi@|
>
>
> notice that bad one starts with ..PNG and the good one with .PNG (0a
> 89 50 vs 89 50)
>
> I've opened it with mc and saw some blank line at the beggining of the
> file. After removing it, the png is displayed again.
>
> What could be insertering that blank line?
in multiple-graphs.php
function mg_pref_order_cmp($a, $b) {
if ($a == $b) return 0;
global $mg_preferred_order;
static $pref_values;
if (!$pref_values) $pref_values = array_values($mg_preferred_order);
$x = array_search($a, $pref_values);
$y = array_search($b, $pref_values);
$found_x = ($x !== FALSE) && ($x !== NULL);
$found_y = ($y !== FALSE) && ($y !== NULL);
if( $found_x && $found_y)
return ($x < $y ? -1 : 1);
if( $found_x)
return -1;
if( $found_y)
return 1;
return $a < $b ? -1 : 1;
}
return (0);
^^^^^^^^^^^^^
that return (0); does the trick...
Thanks to all for the help!
Cheers,
Arnau
|
|
From: Arnau B. <arn...@pi...> - 2009-07-28 08:48:15
|
On Tue, 28 Jul 2009 09:31:13 +0200 Paul Millar wrote: > On Monday 27 July 2009 11:10:39 Arnau Bria wrote: Good morning! > > Glad to see that you're still supporting this project... Last post > > on blog are one year old! > > Sadly true, the blog has suffered lately. The move from the UK to > Germany has soaked up a lot of spare time and, now that my main job > doesn't involve MonAMI, I have less time to spend. > > Still I've some nice things in the pipe-line for MonAMI ... we'll wait for them! [...] > When you got this image on disk, try examining its contents. Perhaps > the easiest is to use hexdump (I believe a pretty standard tool); for > example: > > hexdump -C /bin/bash | head -20 > and > hexdump -C /bin/bash | tail -20 I saw some important diff: bad one: arnau@lx-arnau ~ $ hexdump -C torque-graph.png | head -20 00000000 0a 89 50 4e 47 0d 0a 1a 0a 00 00 00 0d 49 48 44 |..PNG........IHD| 00000010 52 00 00 01 8d 00 00 00 ac 08 06 00 00 00 87 20 |R.............. | 00000020 f0 f6 00 00 00 4a 74 45 58 74 53 6f 66 74 77 61 |.....JtEXtSoftwa| 00000030 72 65 00 52 52 44 74 6f 6f 6c 2c 20 54 6f 62 69 |re.RRDtool, Tobi| 00000040 61 73 20 4f 65 74 69 6b 65 72 20 3c 74 6f 62 69 |as Oetiker <tobi| good one (generated with rrdtool) arnau@lx-arnau ~ $ hexdump -C rrdoutput | head -20 00000000 89 50 4e 47 0d 0a 1a 0a 00 00 00 0d 49 48 44 52 |.PNG........IHDR| 00000010 00 00 02 b9 00 00 00 e9 08 06 00 00 00 20 4f 2f |............. O/| 00000020 6a 00 00 00 4a 74 45 58 74 53 6f 66 74 77 61 72 |j...JtEXtSoftwar| 00000030 65 00 52 52 44 74 6f 6f 6c 2c 20 54 6f 62 69 61 |e.RRDtool, Tobia| 00000040 73 20 4f 65 74 69 6b 65 72 20 3c 74 6f 62 69 40 |s Oetiker <tobi@| notice that bad one starts with ..PNG and the good one with .PNG (0a 89 50 vs 89 50) I've opened it with mc and saw some blank line at the beggining of the file. After removing it, the png is displayed again. What could be insertering that blank line? > > but it's not shown in ganglia... > > > > does it make sense to you? > > Not yet, but I have hope! thanks for your help! > Cheers, > Paul. Cheers, Arnau |
|
From: Paul M. <p.m...@ph...> - 2009-07-28 07:31:54
|
On Monday 27 July 2009 11:10:39 Arnau Bria wrote: > Glad to see that you're still supporting this project... Last post on > blog are one year old! Sadly true, the blog has suffered lately. The move from the UK to Germany has soaked up a lot of spare time and, now that my main job doesn't involve MonAMI, I have less time to spend. Still I've some nice things in the pipe-line for MonAMI ... > Before going to pic problem, I'd like to comment that I solved the php > error by sorting include_once "./functions.php" and include_once > "./get_context.php". Thanks. Good to hear! This should now be fixed in CVS. [..] > before enabling debug, if I click on tho image I get this error: > The image > “http://ganglia-test.pic.es/ganglia/torque-graph.php?g=2&c=Grid%20Services& >h=pbs02.pic.es&r=week&z=large” cannot be displayed, because it contains > errors. It would be useful to have a copy of what the browser is getting: as Andrew says, it might be some PHP error message in-line with the binary data. Could you try saving the broken image's contents from the browser? Perhaps the easiest way (if you're using firefox) is to use the Save As.. button in the Media part of the Page Info (Tools > Page Info, or Ctrl-I). Alternatively, you could use wget to obtain the binary. When you got this image on disk, try examining its contents. Perhaps the easiest is to use hexdump (I believe a pretty standard tool); for example: hexdump -C /bin/bash | head -20 and hexdump -C /bin/bash | tail -20 > start debugging: > [long rrdtool invocation, output redirected to rrdoutput] > > $ file rrdoutput > rrdoutput: PNG image data, 697 x 233, 8-bit/color RGBA, non-interlaced > > then, if I open /tmp/rrdoutput with my browser, I CAN SEE THE IMAGE!! Ah, fun. I'd guess the PHP/Apache somehow have a different environment to your command-line, although I couldn't say what the difference is. You could compare the output hexdump returns with this file against the output from the browser's file. > but it's not shown in ganglia... > > does it make sense to you? Not yet, but I have hope! Cheers, Paul. |
|
From: Andrew E. <and...@gm...> - 2009-07-27 09:37:08
|
> The image “http://ganglia-test.pic.es/ganglia/torque-graph.php?g=2&c=Grid%20Services&h=pbs02.pic.es&r=week&z=large” cannot be displayed, because it contains errors. ... > then, if I open /tmp/rrdoutput with my browser, I CAN SEE THE IMAGE!! > but it's not shown in ganglia... > > does it make sense to you? Yep -- there's probably some extra non-image data being sent to the browser (ie stderr from rrdtool, php warnings etc etc) as part of the <img> call, stupid browsers can't just pick up the "correct" stuff by themselves A |
|
From: Arnau B. <arn...@pi...> - 2009-07-27 09:19:43
|
On Mon, 27 Jul 2009 10:23:00 +0200 Paul Millar wrote: > Hi Arnau, Hi Paul, > On Thursday 23 July 2009 16:03:02 Arnau Bria wrote: > > firs of all I'd like to thank Paul Millar and other contributors for > > your great job on this project. I've been looking for some > > maui/torque graphs and your project really fits our needs. Many > > thanks guys! > > Happy to help! Glad to see that you're still supporting this project... Last post on blog are one year old! Before going to pic problem, I'd like to comment that I solved the php error by sorting include_once "./functions.php" and include_once "./get_context.php". Thanks. [...] > You need to switch on "debug" to see the command that is being run. > Here's how: > > o Edit the file torque-graph.php and change the line: > $rrdgraph->cmd_go( $cmd); > to > $rrdgraph->cmd_go( $cmd, 1); > > o Open the page with your browser and click on one of the > torque graphs. This should view a "large" version of the graph > without any HTML. This allows the browser to respect the MIME type > (which is now text) so the command should be visible. > > So, you should see a page containing the RRDTool command that would > have been run. > > Sometimes the problem is obvious (if one knows how to drive RRDTool), > but if not then try running this command (in a standard console) and > see what error message RRDTool returns. > > Be careful to redirect the output somewhere (e.g., /tmp/rrdoutput) as > RRDTool will emit graphical output (e.g., PNG) to stdout if > everything is OK. (Most terminals don't take too kindly to having > raw binary data on stdout) before enabling debug, if I click on tho image I get this error: The image “http://ganglia-test.pic.es/ganglia/torque-graph.php?g=2&c=Grid%20Services&h=pbs02.pic.es&r=week&z=large” cannot be displayed, because it contains errors. start debugging: # /usr/bin/rrdtool graph - --start -604800 --end N --lower-limit 0 --title 'All jobs by state' --vertical-label 'Jobs' --imgformat PNG --width 600 --height 150 --watermark 'Data collected with MonAMI' DEF:'v0'='/var/lib/ganglia/rrds/Grid Services/pbs02.pic.es/torque.Jobs.State.running.rrd':'sum':AVERAGE AREA:v0#3eb622:'running' DEF:'v1'='/var/lib/ganglia/rrds/Grid Services/pbs02.pic.es/torque.Jobs.State.exiting.rrd':'sum':AVERAGE STACK:v1#9afefd:'exiting' DEF:'v2'='/var/lib/ganglia/rrds/Grid Services/pbs02.pic.es/torque.Jobs.State.completed.rrd':'sum':AVERAGE STACK:v2#bbffbb:'completed' DEF:'v3'='/var/lib/ganglia/rrds/Grid Services/pbs02.pic.es/torque.Jobs.State.held.rrd':'sum':AVERAGE STACK:v3#ff0000:'held' DEF:'v4'='/var/lib/ganglia/rrds/Grid Services/pbs02.pic.es/torque.Jobs.State.suspended.rrd':'sum':AVERAGE STACK:v4#e0e0e0:'suspended' DEF:'v5'='/var/lib/ganglia/rrds/Grid Services/pbs02.pic.es/torque.Jobs.State.transiting.rrd':'sum':AVERAGE STACK:v5#ff7c3d:'transiting' DEF:'v6'='/var/lib/ganglia/rrds/Grid Services/pbs02.pic.es/torque.Jobs.State.waiting.rrd':'sum':AVERAGE STACK:v6#f79afe:'waiting' DEF:'v7'='/var/lib/ganglia/rrds/Grid Services/pbs02.pic.es/torque.Jobs.State.queued.rrd':'sum':AVERAGE STACK:v7#eff360:'queued' > /tmp/rrdoutput $ file rrdoutput rrdoutput: PNG image data, 697 x 233, 8-bit/color RGBA, non-interlaced then, if I open /tmp/rrdoutput with my browser, I CAN SEE THE IMAGE!! but it's not shown in ganglia... does it make sense to you? > Cheers, > > Paul. Thanks for your help Paul! Cheers, Arnau |