I recently updated to teh latest version (6.0) of nimsoft as well as updated all Hubs and Robots and found that in many instances teh net_connect probe will consume 99% of the CPU.has any one else seen this or resolved this?
I have not updated to the latest Nimsoft, however I did just update to the latest net_connect probe. All went well on all of my servers except one that was running Win2003 Server, and I too saw the cpu usage get taken over by net_connect.exe. I don't normally use net_connect on that server so I just deactivated it, but would also like to know what the issue is. I didn't have this problem on any of the other servers running both older and newer versions of the OS when updated to the same version of the net_connect probe.
How many profiles and threads are you running? What sort of config do you have for timouts and retries? If your retries are taking longer than a complete probe cycle to complete it can start to eat resources if there are any errors.
I have all linux hubs and the all saw a cpu spike with the latest version of netconnect maybe not to 90% but all when from say 10 -> 60% or so some had higher. I have felt like submitting a case so I just downgraded to resolve the issue.
I even saw the spike on hubs that only had say 20 profiles and it was a immediate spike soon as probe was deployed cpu started going up.
I only had about a dozen profiles on one and no profiles on the other
to be safe I've deactivated these probes less they interfere with current operations but a better resolution would be prefered
Yea, we definately saw this behavior as well. The newest net_connect probe with no running profiles eats up all the CPU. Haven't gotten around to putting in a case yet, but will do so today after seeing these posts.
The machine that I saw it on didn't have ANY profiles built on it yet. The issue seems to be with the probe itself, not with the number of profiles or the setup of the profiles.
I just put a processes monitor on mine. Latest version running about 70 profiles configured by USM. Nothing very agressive and its using 25% cpu on a 4 core beast. Does anyone have a call open on this one?
I just open a case with nimsoft on this one. What I saw on multiple servers was net_connect segfaulting only happens on this version not sure if they compiled the binaries different or something but 2.73 work like a champ. I also tried upgrading glib and gcc without it having any effect.
net_connect: segfault at 00002aaa00000910 rip 0000003a0b279640 rsp 0000000041350a38 error 4net_connect: segfault at 00000000009f2c30 rip 0000003a0b27341a rsp 0000000043b8a8a0 error 6net_connect: segfault at 00002aaa00aee000 rip 0000003a0b27ba01 rsp 00000000491454c8 error 6net_connect: segfault at 000000000019d870 rip 0000003a0b27341a rsp 00000000493c6ee0 error 6net_connect: segfault at 0000003a00553b98 rip 0000003a0b2733d5 rsp 000000004823eee0 error 4net_connect: segfault at 00002aaa000011f0 rip 0000000000430869 rsp 0000000046326510 error 4net_connect: segfault at 0000000000dc2970 rip 0000003a0b272b3d rsp 0000000041d1c450 error 4
Support has also verified a bug in the new discovery agent no snmpv3 support.
Just found another single core server running Win2008R2 with net_connect.exe consuming 100% of the CPU. And curiously, I have a Quad Core server, also running Win2008R2 with net_connect.exe consuming exactly 25% of each core. Things that make you say, "Hmmmm...."
Support has stated that enginneering is aware of the issue and to resolve downgrade. Next ver of the probe will have the fix.
We have also encounterd this issue with the new net connect probe. Support needs to pull this probe from GA archive.
We've seen this as well - no profiles and 100% CPU usage. I agree that this version of the probe should not be in the archive at this time.
same over here.
I just noticed this yesterday after installing net connect v2.81 on a few linux servers. All had 99% cpu usage with no profiles even created on them. I removed them but have not tried installing 2.71.
I have raised an internal request for this to be removed from the internet archive until a fix is released.
does any one have teh previous version they can attach to this post so I can roll back the update until they getthis resolved
You should still be able to download it from the archive:
This is targeted to be resolved in net_connect 2.9x (no ETA yet).
I have 162 profiles. Downgraded to 2.73 and the CPU load dropped considerably
If you downgrade to 2.73 it should help. Other than that, I have the Nimsoft hotfix for this issue from a case I had opened for it. I can send it if you'd like. I don't know how they feel about attaching directly to the forums. Just send me a quick e-mail at firstname.lastname@example.org
EDIT: The hotfix is for 2.81
Yea I believe version 2.81 fixes the issue but I don't believe they are officially releasing this version but I would think if you ask support you could get it. But the downgrade to 2.73 works as well.
have version 2.81 insatlled and eating-up 50% of cpu
I would HIGHLY recommend rolling back to v2.73 until v2.9x comes out in Q4. There is a new (maybe not so new), more nefarious, bug that is turning off profiles when they fail consecutive ICMP tests. I am told by support that it only affects net_connect v2.8x and that rolling back to 2.73 will alleviate the problem.
If you have any further questions, please don't hesitate to contact me.
net_connect 2.82 is GA now and can be downloaded from the Internet archive. But given the previous comment, I am not so sure I should use it...
I had the same issue with a version of net_connect and proccess probe, the solucion is either to downgrade the version or ask for a hotfix or new release of the probe.
We noticed the same issue with 2.92 so I downgraded to 2.73.
I once had this problem and I applied the hotfix that comes in the following link.
Retrieving data ...