Comware

All my 2650s are running at 20% cpu all the time whether or not there's any traffic.
Firmware H.07.50.
Is this normal or will it be corrected in the upcoming firmwareupgrade?

S.

Well well, our 2626 shows:

12 percent busy, from 1880448 sec ago
1 sec ave: 15 percent busy
5 sec ave: 32 percent busy
1 min ave: 16 percent busy

Same firmware (H.07.50), and no active network traffic at all.

Søren,

We have the same on our 2650's.

We are running RSTP and IGMP filtering, no LACP and high level fault detection. Are you running similar?

Chr

Almost.

RSTP off, LACP off, Qos-passthrough-mode on, IGMP on, fault detection high.

S.

I applied the new H.08.53 firmware to a couple of switches and now they run at 33% with no traffic!!!

Does anyone from HP check these forums?

S.

Hello All -

High CPU utilization is not necessarily a problem. Let's consider a different type of product.

On my Windows PC I sometimes run spybot and ad-aware in the background at low priority at the same time that I am running a browser. Task Manager shows 100% CPU utilization during this time, even though my browser performance is fine.

Of course, I did not call ad-aware or spybot Support to complain about this, since it is not a problem.

So, "high CPU utilization" does not necessarily equal "some process is consuming the CPU, causing problems for other processes". It all depends upon how these processes are prioritized.

The 2650 may have a low-priority background task that is using a lot of the CPU, but this may be per-design and not a problem.

I booted H.08.53 just now, and saw high CPU utilization for less than a minute (I think; I did not actually time it), then the CPU utilization dropped below 10%, even though I was running TELNET and Web to the switch at the same time. The initial high CPU utilization was probably due to some sort of initialization activity.

Obviously, several of you are seeing different results. Perhaps it is due to different configurations.

But, back to the original problem statement. The overall question in my mind is, Are you seeing an actual operational problem?

Ralph

The event log is filled with hundreds of:

Port XX is now off-line

and 2-3 seconds later:

Port XX is now on-line.

This appears randomly for alle ports - and sometimes twice within a minute on the same port.

S.

This could be due to a mismatch on auto/fixed duplex/speed configuration.

If switch port set to auto - end station must be set to auto.

If switch port has a fixed configuration - end station must be set to fixed configuration.

Regards
Regnar

Both ends are set to auto.

PS. Hils Rune :-)

It is rather interesting that you're seeing port toggles. Are you seeing this toggle on the same port or sets of ports? Is it possible that this port toggle is related to the host? Are you seeing this port toggle issue on other 2600's? This may simply be bad hardware.

I just checked the web and noticed that the latest code for the 2600 is H.08.55. You *may* want to see if this resolves your problem.

Good luck.

S

As for the port toggling - this looks like it could be LACP checking for trunk partners (or whatever it is called) when a machine brings up the port.

I have seen this in my logs as well... it's not a problem in essence... just fills the logs :)

Regards,
Shane.

Newest firmware is H.08.53.

I think this is happening on all ports - I have definitely seen it on most ports.
Nics connected are RTL 8139 and Intel integrated and others.

At first I thought that it was due to LACP or STP (RSTP) so I *disabled* both, but it didn't make any difference.

We're seeing random problems with saving files to UNC drives and mapped drives disappearing which might be due to the network connection fluctuating?

S.

Hi Søren,

In general the CPU Util of the switch should not be affected by normal traffic. Your level of high util, could be caused by broadcasts on the network. Other things also consume CPU time, internal logging, statistics update, response to SNMP packets (if that's how you measure it), and maintainance of routing tables etc.

In general my experience is that ~ 15% is normal...

Held og lykke ;)

Uf