Problem with Cactid under Solaris

[Ev]

I configure Cactid under Solaris 9.

[root@o] (/var/home/cacti/source/cacti-cactid-0.8.6d)> ./configure --prefix=/opt/cactid --with-mysql=/var/home/cacti/source/mysql-4.1.9/include --with-snmp=/var/home/cacti/source/net-snmp-5.2.1.rc3/include/net-snmp --with-openssl=/var/home/cacti/source/openssl-0.9.7e

Mysql sources is present, but :

checking for strtoll... yes
configure: error: Cannot find MySQL header files under /var/home/cacti/source/mysql-4.1.9/include

Why ?

gcc version 3.4.2
mysql v4.1.9
net-snmp 5.1.2-rc3

[zanaguara]

Try:

--with-mysql=/var/home/cacti/source/mysql-4.1.9

[rony]

I was just about to suggest the same. Try that and let us know what happens.

[Ev]

# path to .lib & header directiries on installed mysql & net-snmp, not to # sources.

./configure --prefix=/opt/cactid --with-mysql=/opt/mysql --with-snmp=/opt/net-snmp

# path to library
#LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib:/opt/net-snmp/lib:/opt/mysql/lib
# /mysql

make & make install

normal complete.

Cactid is faster - 4* poleer.php is 210..230 sec (62 hosts), 4*cactid*4 threads - 150...160 sec, but much less memory & cpu load.
solaris 9, 2*UltraSparcIIe 440Mhz , heavy load under another program.

But Cactid processes hang and remain in memory - it is necessary to remove them on kill-9 - it don't ended
debug:

01/27/2005 03:35:02 PM - POLLER: Poller[0] Maximum runtime of 296 seconds exceed
ed. Exiting.
01/27/2005 03:35:04 PM - POLLER: Poller[0] DEBUG: About to Spawn a Remote Proces
s [CMD: /opt/cactid/bin/cactid, ARGS: 0 21]
01/27/2005 03:35:04 PM - POLLER: Poller[0] DEBUG: About to Spawn a Remote Proces
s [CMD: /opt/cactid/bin/cactid, ARGS: 22 41

- but process is remain in memory. 8-(

What I can send to you (another debug?) for correct this bug ?

[Ev]

problem is partially located - cactid processes was remain in memory, when it starts from cron (from user cacti) and (as test - it runs in parallel, from user cacti) once from script.

cacti sometimes don't kill this cactid daemons - please, correct this.
on example , some old processes (start at 11:57) was remain in memory.

[@] (~)> ps -ef | grep cacti
cacti 5946 1 0 12:00:03 ? 0:00 /opt/cactid/bin/cactid 23 42
cacti 6022 5946 0 12:00:05 ? 0:00 /opt/php/bin/php /var/home/cacti/cacti-0.8.6c/script_server.php cactid 0
cacti 5781 5773 0 11:57:58 ? 0:00 /opt/php/bin/php /var/home/cacti/cacti-0.8.6c/script_server.php cactid 0
cacti 7572 156 0 12:05:01 ? 0:00 sh -c . /var/home/cacti/.bash_profile; /opt/php/bin/php /var/home/cacti/cacti-0
cacti 7741 1 0 12:05:05 ? 0:00 /opt/cactid/bin/cactid 0 22
cacti 7772 7771 0 12:05:06 ? 0:00 /opt/rrdtool/bin/rrdtool -
ev 9149 17969 0 12:09:02 pts/5 0:00 grep cacti
cacti 7613 7572 0 12:05:01 ? 0:02 /opt/php/bin/php /var/home/cacti/cacti-0.8.6c/poller.php
cacti 5773 1 0 11:57:58 ? 0:00 /opt/cactid/bin/cactid 43 65
cacti 7771 7613 0 12:05:06 ? 0:00 sh -c /opt/rrdtool/bin/rrdtool -
cacti 7785 7741 0 12:05:06 ? 0:00 /opt/php/bin/php /var/home/cacti/cacti-0.8.6c/script_server.php cactid 0

[Ev]

these problems was founded, when number of cactid processes was 2 or 4 (and 2 - 4 threads per process).

[Ev]

if in moment, when cactid works, change configuration (on example - add new host ) , Cactid process (run 1 process * 4-5 threads) was remain in memory and don't stop by cacty.

[Ev]

Cactid 0.8.6.e : this problem is present.

Can anyone wrote about it's cactid under Solaris ?

[Ev]

Now I partilly solved this problem with external kill of the cactid process -

sudo crontab -l cacti
# For cacti
0,5,10,15,20,25,30,35,40,45,50,55 * * * * . /var/home/cacti/.bash_profile; /opt/php/bin/php /var/home/cacti/cacti-0.8.6c/poller.php
# kill cactid
2,7,12,17,22,27,32,37,42,47,52,57 * * * * /bin/pkill cactid

# cactid polled all hosts at 30...40 sec.
# if cactid is in memory after 120 sec of work - it is hang -->> kill it

isn't good solution , but it working.

[TheWitness]

Cactid 0.8.6e was release prematurely. That release is not ready for production. Please downgrade and then lets talk some more.

In the mean time, please get off of the RC3 of Net-SNMP to a production stable version.

TheWitness

[Ev]

versions 0.8.6d & 0.8.6e - Similar problems.

[TheWitness]

Ev,

Have your upgraded/downgraded net-snmp?

TheWitness

[Ev]

Yes.
But last problem polling cycle (1.5 hours ago) was with prev. version (5.1.2-rc3) of net-snmp.
I wrote results with 5.1.2 later.

[Ev]

with installed net-snmp 5.1.2 - similar problem .

[Ev]

Under which OS you run cactid ? (Linux ? Solaris ? )

We have similar problem under Solaris 9