Page 8 of 9

Re: Memory usage

Posted: Tue Dec 24, 2019 7:43 pm
by Ludvik
@jorgenekah: I do not have STP enabled on any of my netonix switch.

2019-12-25_year.png

2019-12-25_week.png

Re: Memory usage

Posted: Thu Jan 02, 2020 10:34 pm
by Ludvik
Maybe you still have problem with discovery:
after_disable_discovery-month.png

I uncheck all of checkboxes. Version 1.5.5

We will after a few days to see if it helped.

Re: Memory usage

Posted: Fri Jan 03, 2020 2:57 pm
by Stephen
Possibly, keep us informed.

Re: Memory usage

Posted: Sat Jan 11, 2020 8:13 am
by Ludvik
Small detail, but significant. On friday I enabled all of discovery checkboxes.
10_106_30_74__1_3_6_1_4_1_2021_4_6_0-week.png

10_106_30_74__1_3_6_1_4_1_2021_4_6_0-month.png


10_106_30_74__202001180208-month.png

Re: Memory usage

Posted: Fri Feb 07, 2020 10:15 pm
by Ludvik
Any news? Thanks.

Re: Memory usage

Posted: Sat Apr 04, 2020 11:21 pm
by Ludvik
Any news?

Re: Memory usage

Posted: Fri Apr 10, 2020 8:04 am
by Stephen
Hey Ludvick, so first thanks for dealing with this for so long. I may have found an insight. So I'm going to run some test's soon. Although for the WS right now I'm trying to get the present issue's with Multicast finished first. We've talked about it before on this thread but I think the issue may be rooted in SNMP. Thinking about the evidence, SNMP itself is a subsystem that branches it's way into every active protocol on the switch, including discovery, STP, etc. If you remember, we already solved an issue regarding the OID .1.3.6.1.2.1.17.2.5.0 that slightly improved the memory bug. At that time that's what I thought the issue was and I was baffled that it's resolution didn't seem to affect the problem. However, I think I remember another report regarding snmp where certain OID's where not reachable until they where cached via snmp-walk or querying certain more directly accessible OID's. The implication here is a that a number of memory management related task's are being done that might not be being handled correctly.

So what I'm doing is I have a script that is performing snmp-walk at very short intervals continuously. I'm going to watch that for awhile to see what it does to a switch running 1.5.5

If possible, could you note down what the memory usage is on one of your switches and then disable snmp on on it for a few days. Then re-enable snmp and query the memory. If I'm right, it should not have grown in the time-span that snmpd was disabled.

That would be very helpful to know.

Re: Memory usage

Posted: Fri Apr 10, 2020 7:20 pm
by Ludvik
OK. Will see after a week.

Before disabled SNMP: used RAM 56.3. Immediately after disabled: 55.8

Re: Memory usage

Posted: Wed Apr 15, 2020 9:50 pm
by Ludvik
Used RAM is still growing.

11.4.2020 55.8
12.4.2020 1:15 56.8
14.4.2020 3:20 59.0
16.4.2020 3:47 61.4
20.4.2020 1:47 66.5
23.4.2020 1:15 69.7
28.4.2020 1:00 75.0
01.5.2020 3:00 78.3
03.5.2020 5:15 80.6
07.5.2020 0:46 84.7
09.5.2020 1:20 86.8
20.5.2020 22:45 99.3

I think ... SNMP is not problem.

Re: Memory usage

Posted: Mon May 04, 2020 7:31 pm
by rclark
I just noticed today that one of our switches is using 112MB of memory. We just started using librenms so I only have a few days of graph.