View Full Version : CPU usage is between 80% -90% after Mgmt Sevre upgrade

2013-01-18, 13:12
Hi all,

hope someone could spread some light on this.
problem is, ever since upgrading the mgmt Server from R71.4 to R75.4
the edge device (Firewalls) have increase on CPU usage. but the Edge devices were not
upgrade to R75.40 they are still on R71.4 (on IPSO 6.2-GA055b06).

the only thing that has changed before the high CPU was the upgrade to R75.4 on the mgmt Server, and from the Voyager history statistics we don't see an increase in number of packets or bytes on the interfaces if compared to the week before.

What was seen was, high system usage (often >80%), main process which was causing this was net_taskq0. Connections were between 12000 and 20000, the current peak value is 21262:

xxxxx[xxxx]# fw tab -t connections -s
localhost connections 8158 3568 21262 10696
On the interfaces I see a high number of drops under RxNoBuffer, which seems that the system cannot empty the interface buffers quick enough.

Any advice/help would be greatly appreciated
Thanks in advance

2013-01-25, 12:49
I have never used a Nokia/Voayager setup - only SPLAT and GAIA; assume its the same.

Are you running the mangement on an appliance or a VM? can you post a top command to confirm whats using the CPU

I recently upgraded the a MDS to EA and had the same issue - it was beacuse the VM was running on SATA disks and the intel team saturated its usage. I only noticed the degregation in speed when I restarted server and all MDS instances where starting up. Resolved issues by moving VM to less utalised SAS disks.

2013-01-28, 09:33
Hi ra3ac

thanks for the reply, however this issue was resolved :)
after running zdebug we were able to see two host generating lots of traffic.
after removing them from the network CPU usage dropped to 40% (but this is still more CPU usage then before upgrade(by 10%))