There's a new release-candidate available at http://www.hswn.dk/beta/
Changes from RC1 include:
Fixed lockup when hobbitd loaded a checkpoint file containing scheduled disable-requests.
The [disk1] graph definition was broken.
Using the "apache" test on a host with no HTTP tests now works, and does not result in a bogus http column.
The bbmessage CGI was using a broken method to fetch the POST data.
hobbit-enadis could crash.
Some AIX installations apparently do not have atoll(). Use the Hobbit internal atoll() routine on AIX.
The "apache" RRD handler now recognizes reports from Apache 1.x for Busy/Idle server processes.
The "sendmail" RRD handler will create RRD-files with room for the "msgsqur" (quarantined messages count) reported by sendmail 8.13+. Existing RRD files will continue to work, but will not track this data item.
A new reporting option lets you save the availability data in a comma-separated file, for import to spreadsheets.
Updates of RRD files now use the RRDtool "template" facility to map data-values to the correct RRD datasets. This should solve the problem with migrating of vmstat data from several systems, including AIX.
A couple of log files have moved to /var/log/hobbit: The nkstatus.log file (from the "server/" directory); and the acklog and notifications.log files (from the data/acks directory).
Hobbit now works when compiled with RRDtool 1.2.x.
/lib64 is searched for libraries by the configure script.
You can now define environment "areas", i.e. an environment variable defined in hobbitserver.cfg can have different values. This lets you eg. setup the BBDISP variable with different values for the maintenance tools, so you can send "disable" messages also to your standby Hobbit server.
You can limit utilisation tracking of filesystems via the new RRDDISKS and NORRDDISKS environment settings.
A new "contrib" directory was added to the distribution. It currently includes a couple of Hobbit logo GIF's, and a HOWTO document describing how to setup a clustered Hobbit installation.
Henrik
Hi,
while running configure I got: collect2: ld returned 1 exit status make: *** Ätest-linkÅ Error 1 WARNING: LDAP library files found in /usr/local/lib, but link fails. ./configureÄ120Å: ./test-ldap: not found.
Regards Lars Hobbithobbyist
Http://leopg9.no-ip.org ----- Original Message ----- From: "Henrik Stoerner" <henrik at hswn.dk> To: <hobbit at hswn.dk> Sent: Sunday, May 08, 2005 11:27 PM Subject: [hobbit] Hobbit 4.0.3rc2 available
There's a new release-candidate available at http://www.hswn.dk/beta/
Henrik Stoerner wrote:
There's a new release-candidate available at http://www.hswn.dk/beta/
Yay, glad to get my hands on this one. ;)
- Using the "apache" test on a host with no HTTP tests now works, and does not result in a bogus http column.
I can confirm that this works now. Thanks.
- A new reporting option lets you save the availability data in a comma-separated file, for import to spreadsheets.
Is this cgi-bi/bb-csvinfo.sh? I have to manually edit the menu to add it, huh?
- Updates of RRD files now use the RRDtool "template" facility to map data-values to the correct RRD datasets. This should solve the problem with migrating of vmstat data from several systems, including AIX.
If I understand this correctly, this could help me get rid of 2 local patches that I maintain to keep RRD data consistent between my inconsistent legacy systems. Is its use transparent or something I can tap into? Right now I have to force some column re-ordering in a couple of the hobbitd/larrd/do_xxx routines to make things work, and editing the source files has been the easiest & most consistent way to do it.
- A couple of log files have moved to /var/log/hobbit: The nkstatus.log file (from the "server/" directory); and the acklog and notifications.log files (from the data/acks directory).
FYI to users going from older hobbit, you may want to copy over the contents of these files into these new locations, otherwise you could lose a lot of "history" in these areas (not to mention any current acks that are in effect when you apply the RC2 upgrade).
Tom
On Mon, May 09, 2005 at 10:37:16AM -0400, Tom Georgoulias wrote:
Henrik Stoerner wrote:
- A new reporting option lets you save the availability data in a comma-separated file, for import to spreadsheets.
Is this cgi-bi/bb-csvinfo.sh? I have to manually edit the menu to add it, huh?
No, it's currently only available by running bbgen with the --csv option in reporting mode. I'll add something to the report-webpage so you can get the CSV output directly.
- Updates of RRD files now use the RRDtool "template" facility to map data-values to the correct RRD datasets. This should solve the problem with migrating of vmstat data from several systems, including AIX.
If I understand this correctly, this could help me get rid of 2 local patches that I maintain to keep RRD data consistent between my inconsistent legacy systems. Is its use transparent or something I can tap into? Right now I have to force some column re-ordering in a couple of the hobbitd/larrd/do_xxx routines to make things work, and editing the source files has been the easiest & most consistent way to do it.
Yes, this was done solely for you :-)
It should happen automatically.
- A couple of log files have moved to /var/log/hobbit: The nkstatus.log file (from the "server/" directory); and the acklog and notifications.log files (from the data/acks directory).
FYI to users going from older hobbit, you may want to copy over the contents of these files into these new locations, otherwise you could lose a lot of "history" in these areas (not to mention any current acks that are in effect when you apply the RC2 upgrade).
The acks shouldn't be affected, they are maintained by hobbitd in it's own checkpoint file. At least if you upgrade from one Hobbit version to another (not from BB to Hobbit).
Regards, Henrik
Henrik Stoerner wrote:
The acks shouldn't be affected, they are maintained by hobbitd in it's own checkpoint file. At least if you upgrade from one Hobbit version to another (not from BB to Hobbit).
It would probably be better to write that I lost the list of events ack'd within 240 minutes on the non-green display page. Sorry for the confusion.
Tom
I just noticed that after I upgraded to rc2, selecting the enable/disable tool from the menu when viewing a group of systems only allows me to enadis those systems in the group that I was just viewing. The only way to see all systems on my network is to go back to the main view, then use the enadis tool. (Or I can cheat and click "Apply filters" in the Filter Hostlist box while leaving all the search fields empty).
How can I revert back to the old way, where selecting enable/disable from the menu will display all the systems? I can see how this new behavior is useful, but I've also witnessed first hand how it can be confusing as well.
Tom
On Mon, May 09, 2005 at 02:25:26PM -0400, Tom Georgoulias wrote:
I just noticed that after I upgraded to rc2, selecting the enable/disable tool from the menu when viewing a group of systems only allows me to enadis those systems in the group that I was just viewing. The only way to see all systems on my network is to go back to the main view, then use the enadis tool. (Or I can cheat and click "Apply filters" in the Filter Hostlist box while leaving all the search fields empty).
That's not cheating - that's the way I wanted it to work.
"It's a feature - not a bug" as every software developer says when faced with two incompatible requests.
In my setup, the full list of hosts is about 2000 systems. You really don't want to pick hosts from a selection list that size ...
How can I revert back to the old way, where selecting enable/disable from the menu will display all the systems?
You can modify the hobbit-enadis.sh wrapper script and unset the HTTP_COOKIE variable that is used to track what page the request comes from. Put HTTP_COOKIE= export HTTP_COOKIE before the "exec /..../hobbit-enadis.cgi" line.
I can see how this new behavior is useful, but I've also witnessed first hand how it can be confusing as well.
Sorry ... but I think the confusion happens only because you've been used to the way the maint.pl script works.
Regards, Henrik
[Mon May 09 19:50:50 2005] [error] [client x.x.x.x] Premature end of script headers: hobbit-enadis.sh, referer: http://hobbit.domain.com/hobbit-seccgi/hobbit-enadis.sh
I cp'ed the rc1 cgi binary over the rc2 cgi binary
hobbit at hobbit ~/server/bin> pwd /export/home/hobbit/server/bin
hobbit at hobbit ~/server/bin> cp /var/tmp/hobbit-4.0.3rc1/hobbitd/hobbit-enadis.cgi .
Ideas? It works fine with the rc1 binary.
--
David
On Mon, May 09, 2005 at 08:10:34PM +0000, David Gore wrote:
[Mon May 09 19:50:50 2005] [error] [client x.x.x.x] Premature end of script headers: hobbit-enadis.sh, referer: http://hobbit.domain.com/hobbit-seccgi/hobbit-enadis.sh
It's a bug - I copied a couple of lines and missed changing them to fit the new use.
Patch at http://www.hswn.dk/beta/rc2-enadis.patch
Henrik
participants (4)
-
David.Gore＠mci.com
-
henrik＠hswn.dk
-
lars.ebeling＠leopg9.no-ip.org
-
tgeorgoulias＠mcclatchy.com