September 2006 - Xymon - Xymon mailman web

Bug in rrd/do_disk.c while using /GROUP option of bb command
by fduranti＠q8.it 30 Sep '06

30 Sep '06

Hi Henrik, I got some problems with RRD creation using the "/group:groupname" directive. The do_disk.c will skip every line that don't contain a "/" and the string " red " or " yellow ". This will work if the test are in error but it will not work if the test are not red or yellow. The first line of the message sent with the group directive is something like status+60/GROUP:netappadmin VITROMFS65.disk green Sat Sep 30 16:59:17 2006 And this will be carried on generating a rrd file with a strange name. This can be avoided skipping the first like changing line 54 of hobbitd/rrd/do_disk.c from curline = msg; To curline = strchr(msg,'\n'); if (curline) curline++; Just to skip the first line. Francesco

1 0

Feature request: "more" on trends columns
by fduranti＠q8.it 30 Sep '06

30 Sep '06

Well just to put some more on new feature that could be useful... I'm currently getting some more data to graph for netapp filers via the stats command. >From that command I'm getting a big amount of data: interfaces, per volume/aggregate/qtree statistics, single disk utilization statistics. >From all those data I'll end up with a really big number of possible graphs and putting them all in the trends column will result in an almost unreadable screen (it's difficult to search for a single graph in more then 200). One simple to implement solution (at least to find the graph) could be to create link a the start of the page for each section and, a header before the graphs saying what they are and a footer under the graph to come back to the start of the page. For example: CPU -> CPU stat graph Vmstat -> vmstat graph Interface -> interface graph Disk -> disk graph Snapshot -> snapshot graph Cpu statistics: Graph1 Back to start VM Statistics: Graph1 Graph2 Graph3 Back to start Interface statistics Graph1 Graph2 Graph3 Back to start ... And so on Another nice addiction can be to have more trend columns where user can define what data column will be displayed. In this case we can differentiate graphs by data type. We can have a trend column for disk spaces (disk space, inode, snapshot and so on), a trend column for machine performance (cpu, vmstat, iostat), a trend column for network performance (traffic and tcp statistics) and a trend column for environment status (power, temp, etc). Having the 2 addiction at the same time will get probably the most out the graphs. Francesco

1 0

FILE check - value too large
by davidstuffle＠yahoo.com 30 Sep '06

30 Sep '06

The FILE check is giving the "Value too large for defined data type" error message for files over 2GB. Wasn't their a fix for this? I though it was in the 4.2.0 release, but I'm still getting the error. David --------------------------------- Do you Yahoo!? Everyone is raving about the all-new Yahoo! Mail.

1 0

bb3.html possible?
by rower.master＠gmail.com 29 Sep '06

29 Sep '06

Henrik, Is possible to generate a bb3.html page only for a group of defined hosts using bbgen? \ Thanks in advance. Mario.

2 1

Is "Current non-green Systems" timeframe configurable?
by jonescr＠cisco.com 29 Sep '06

29 Sep '06

I'm being asked how long the time period is for e historical events in the "All non-green view" page is, and well, I'm just not sure. I have seen it say "No events received in the last 240 minutes", so I take that to mean that the history only shows the last 4 hours worth of events. That begs the question of, can that value be changed, and if so, where? -Charles

2 1

title showing status counts?
by mikee＠mikee.ath.cx 29 Sep '06

29 Sep '06

Does anyone have a patch where the HTML title shows the counts for each status? Where the html page would be: <html><title>red - 1:4:0:0:1</title> .... </html> The first numbers is red, the second yellow, the third green, then purple, and finally blue. Mike

1 0

LOG test fails to IGNORE ?
by jjj863＠gmail.com 29 Sep '06

29 Sep '06

hi, I have one single Hobbit server running 4.2RC1-20060712 on CentOS 4. Clients have some harmless SELinux warning messages in /var/log/messages. I configured Hobbit to ignore such lines (conf entries appended). However, the LOG test still generates alerts. Below is excerpted from an email alert. Any idea why? I brought up before 4.2 final version is released, I believe, and was told it should not be the case. &yellow Sep 28 15:58:20 saturn su[4070]: Warning! Could not relabel with user_u:object_r:devpts_t, not relabeling. ---cut---x------ hobbit-clients.cfg -----x-------cut----- HOST=saturn log /var/log/messages %WARNING|NOTICE|ERROR COLOR=yellow IGNORE=relabeling ---cut---x------ clients-local.cfg -----x-------cut----- [saturn] log:/var/log/messages:10240 ignore MARK

2 3

Patch for local clients
by jason.kruse＠teldta.com 29 Sep '06

29 Sep '06

Noticed this one today. Clients configured for local monitoring get "Failed to get a message, terminating" messages throughout their logs. Patch for hobbitd_client.c to check for localmode for this error. 1716c1716 < errprintf("Failed to get a message, terminating\n"); --- > if (!localmode) errprintf("Failed to get a message, terminating\n"); Jason

1 0

[hobbit] hsfs in hobbitclient-sunos.sh
by fduranti＠q8.it 29 Sep '06

29 Sep '06

Well my problem was on nfs mount that didn't get mounted at all at boot. If I'm not wrong at least on linux machine df will hung but the mount command will not hung but it will show the nfs mount and you don't get any warning. > -----Original Message----- > From: David Gore [mailto:David.Gore at VerizonBusiness.com] > Sent: Friday, September 29, 2006 5:48 PM > To: hobbit at hswn.dk > Subject: Re: [hobbit] hsfs in hobbitclient-sunos.sh > > Just keep in mind that an nfs mount that is not working will > hang the client, and you will get no reports, although you > will get purples that you may not see in the middle of the night. > > ~David > > > Francesco Duranti wrote: > > It should be good also for nfs mount ... We had problem in the past > > with some oracle filesystem not getting mounted by the > machine and it > > could be nice to be sure they're mounted > > > > Francesco > > > > > >> -----Original Message----- > >> From: Henrik Stoerner [mailto:henrik at hswn.dk] > >> Sent: Friday, September 29, 2006 5:22 PM > >> To: hobbit at hswn.dk > >> Subject: Re: [hobbit] hsfs in hobbitclient-sunos.sh > >> > >> On Fri, Sep 29, 2006 at 10:01:56AM -0500, Kruse, Jason K. wrote: > >>> Perhaps a change to hobbitd_client might be more > >> appropriate to handle > >>> this type of media. Something that lets us know a CD is > in the box > >>> yet doesn't send data points for the RRD's. > >> I'm think about having the client handle the output from > "mount" to > >> check for cd's, USB-dongles etc. that you may want to react on. > >> > >> The current clients already report the "mount" data, so > it's purely a > >> server-side change. > >> > >> > >> Regards, > >> Henrik > >> > >> > >> To unsubscribe from the hobbit list, send an e-mail to > >> hobbit-unsubscribe at hswn.dk > >> > >> > >> > > > > To unsubscribe from the hobbit list, send an e-mail to > > hobbit-unsubscribe at hswn.dk > > > > > > > To unsubscribe from the hobbit list, send an e-mail to > hobbit-unsubscribe at hswn.dk > > >

1 0

hsfs in hobbitclient-sunos.sh
by Dominique.Frise＠unil.ch 29 Sep '06

29 Sep '06

Hi, The FSTYPES in hobbitclient-sunos.sh does not exclude cdrom file systems (hsfs). To correct: [bb at iris bin]$ diff -u hobbitclient-sunos.sh hobbitclient-sunos.sh.dist --- hobbitclient-sunos.sh Thu Sep 28 12:07:13 2006 +++ hobbitclient-sunos.sh.dist Sat Aug 19 10:20:00 2006 @@ -22,7 +22,7 @@ echo "[df]" # All of this because Solaris df cannot show multiple fs-types, or exclude certain fs types. -FSTYPES=`/bin/df -n -l|awk '{print $3}'|egrep -v "^proc|^fd|^mntfs|^ctfs|^devfs|^objfs|^nfs|^lofs|^hsfs"|sort|uniq` +FSTYPES=`/bin/df -n -l|awk '{print $3}'|egrep -v "^proc|^fd|^mntfs|^ctfs|^devfs|^objfs|^nfs|^lofs"|sort|uniq` if test "$FSTYPES" = ""; then FSTYPES="ufs"; fi set $FSTYPES /bin/df -F $1 -k | grep -v " /var/run" Dominique UNIL - University of Lausanne

4 3