Page 2 of 4
Re: NCPA reading memory wrong for terabite
Posted: Thu Dec 26, 2019 10:46 am
by mbellerue
Actually, I'm looking over the numbers again, and for the most part they actually look like they're in line with what the server is showing. The biggest variant being the "used" memory on NCPA is showing 4743.91GiB, where as free -h shows 5.9TB. The difference there might be in the conversion from gibibytes to gigabytes to terabytes.
Re: NCPA reading memory wrong for terabite
Posted: Fri Dec 27, 2019 12:28 pm
by rnjie
its not calculating the cache memory because if it did it would not report memory is 97% full when we still have about 1.4T cached
Re: NCPA reading memory wrong for terabite
Posted: Fri Dec 27, 2019 1:36 pm
by mbellerue
Ah, I see, you're absolutely right. Can you modify your check command to add the -n switch to the check? It would go in the -a list of switches to send to the remote system. Or if you can post your command check, I can modify it.
Re: NCPA reading memory wrong for terabite
Posted: Fri Dec 27, 2019 3:09 pm
by rnjie
i added -n i didnt see any changed see command below
check_ncpa.py -H lxdodbt1 -t 'plut0' -P 5693 -M memory/virtual -u Gi -n -w 85 -c 95
Re: NCPA reading memory wrong for terabite
Posted: Fri Dec 27, 2019 3:57 pm
by mbellerue
My apologies, that switch was for check_nrpe, not ncpa. This might just be a bug with the older version of RHEL. I will spin up a RHEL 6 environment on my side and test it out.
Re: NCPA reading memory wrong for terabite
Posted: Thu Jan 02, 2020 4:06 pm
by rnjie
any update on this? happy new year!
Re: NCPA reading memory wrong for terabite
Posted: Thu Jan 02, 2020 4:45 pm
by mbellerue
Happy new year!
I did get back to this. On a fresh CentOS 6 install, with NCPA 2.2.0, I'm getting a used memory result that does not include buffers or cache.
The only thing I can think of is possibly a difference in Python version on the monitored system? If I do python --version, I'm running Python 2.6.6.
Do you have any other RHEL6 systems running NCPA?
Re: NCPA reading memory wrong for terabite
Posted: Mon Jan 06, 2020 10:00 am
by rnjie
yes i do, it has almost thesame memory usage as the other one but its not giving me 98% full, see below
t1 ~]# free -h
total used free shared buffers cached
Mem: 5.9T 5.9T 16G 1.1T 1.1G 1.2T
r4 ~]$ free -h
total used free shared buffers cached
Mem: 5.9T 5.6T 352G 635M 963M 1.0T
t1 is showing 98% full while r4 is showing 75%full, as you can see their usage is not different and they are both rhel6 and monitored by thesame version of ncpa
the command used for monitoring both is thesame
check_xi_ncpa -t 'plut0' -P 5693 -M memory/virtual -u Gi -w 85 -c 95
Re: NCPA reading memory wrong for terabite
Posted: Mon Jan 06, 2020 12:57 pm
by mbellerue
So there must be something different about t1. Can you run this command on both t1 and t4, and give me the output?
Code: Select all
sha256sum /usr/local/ncpa/psutil._psutil_linux.so
Re: NCPA reading memory wrong for terabite
Posted: Mon Jan 06, 2020 2:10 pm
by rnjie
see below
t1 ~]# sha256sum /usr/local/ncpa/psutil._psutil_linux.so
445a6fa47fc9cf4b0769a0959e3dcb37c6f89c7ae3363aefcbe0befb84acf81d /usr/local/ncpa/psutil._psutil_linux.so
r4 ~]# sha256sum /usr/local/ncpa/psutil._psutil_linux.so
4a67db04cb8848d4ad0f2570120b613989e8c268ed2e59ec1b842071f7b987f7 /usr/local/ncpa/psutil._psutil_linux.so