Switches, Hubs, and Modems
1748157 Members
4247 Online
108758 Solutions
New Discussion

Re: 00075 system: Out of pkt buffers; miss count: 1

 
James Crosland_1
Occasional Advisor

00075 system: Out of pkt buffers; miss count: 1

We have recently set up a new corporate LAN with two Procurve
5406zl switches acting as the core and 8 stacks each with three
Procurve 2610-24 switches.

Each stack of end user switches has 2 VLANs, one for voice, one
for data. The core also has separate VLANs for voice and data,
along with a Switch vlan

The two central switches are using VRRP to determine the active
gateway address for each of the vlans, with the "Data" VLANs
going to the "Primary" core switch and the "Voice" VLANs goi
ng
to the secondary core switch.

The LAN has be live for 2 1/2 weeks and last night the secondary
core switch failed for the second time.

The errors appear to start with

W 02/28/12 18:35:48 00075 system: Out of pkt buffers; miss count: 1

after which there are repeated entries as below

E 02/28/12 18:38:26 00785 vrrp: Failed to allocate pkt buffer for VRRP pkt(1)

after 12 minutes we then got these messages relating to NTP failures, I assume because the swithc couldn't access the vrrp
gateway.

I 02/28/12 18:50:51 00411 SNTP: Timed out on server 192.168.48.5.
I 02/28/12 18:50:51 02631 SNTP: Server not found at 192.168.48.5.
I 02/28/12 18:50:51 00411 SNTP: Timed out on server 192.168.48.1.
I 02/28/12 18:50:51 02631 SNTP: Server not found at 192.168.48.1.
I 02/28/12 18:50:51 00414 SNTP: Unable to reach configured SNTP servers

Reloading the switch seemed to resolve the problem, but during the outage our VOIP system stops working - its primary card is plugged in to the secondary (failing) switch as as the switch itself never dies the PABX continues to try using that nic and
therefore all calls stop.

All assistance would be greatly appreciated.

6 REPLIES 6
jguse
HPE Pro

Re: 00075 system: Out of pkt buffers; miss count: 1

Hello,

 

It's hard to determine right now what could be causing the issue, but your packet buffers are overloaded as the errors seem to indicate. Is the switch able to handle the amount of traffic going over the ports? Overloaded switch could be one cause, hardware issue another, or maybe you're hitting some odd unknown software bug.

 

Either way I'd recommend that you prepare a "show tech all" from both switches, logs at the time of the issue if possible, and open a case with support so that we can have a proper look at this.

Best regards,
Justin

Working @ HPE
Accept or Kudo
John Gelten
Regular Advisor

Re: 00075 system: Out of pkt buffers; miss count: 1

A customer of ours has had the same issue two years back, with the same combination of devices, roughly the same data/voice VLAN setup, the same VRRP setup. And most important the same issues of failing switches, not all at the same time. So I am hurting my brain to remember what solved this... We had several issue in that network, so not sure what solution was for what issue...  ;-)

 

My best guess at the moment is that we were using ICMP redirects, which caused a lot of traffic initiated by the switch itself. Resulting in that pktbuffer issue, and the management-part of the switch just stops communicating, so your VRRP and ARP hits the timeout and your network gets really silent...

I will go through my archive to see if I can find a better answer...

 

You are using a recent software image ?

joholmst
Occasional Visitor

Re: 00075 system: Out of pkt buffers; miss count: 1

Hi,

 

We have the same problem with an hp4208vl switch and the syetm went down by itself a few minutes after that (don't know if it's the reason why it went down?).

 

Have you found any solution to this or reason why it does it?

 

 

Arimo
Respected Contributor

Re: 00075 system: Out of pkt buffers; miss count: 1

Hi

 

As this is a new installation, I'd suggest you go to https://h10145.www1.hp.com/downloads/SoftwareReleases.aspx?ProductNumber=J9532A and upgrade the switch SW before anything else. I suspect you're running an old code.

 

Joholmst, go to CLI and give command "show boot-history". This gives you a three-line crash message, which shows the cause of the crash.


HTH,

Arimo
HPE Networking Engineer
joholmst
Occasional Visitor

Re: 00075 system: Out of pkt buffers; miss count: 1

Ok, thanks...and the message is...

 

SubSystem 0 went down:  08/19/12 16:49:35
Software exception at fcDmaUtil.c:587 -- in 'eDMAEmg001', task ID = 0x361b378
-> internal error

Arimo
Respected Contributor

Re: 00075 system: Out of pkt buffers; miss count: 1

Hi

 

If you didn't upgrade yet, do so. If you did and haven't seen the crash since, likely there's nothing to worry about. if you see the crash again, collect the following from the switch:

 

- show tech all -command output

- crash log

- crash data

- core dump

- fdr log

 

copy command-output "show tech all" tftp <server IP> showtech.txt

copy crash-log mm tftp <server IP> crashlog

copy crash-data mm tftp <server IP> crashdata

copy fdr-log all tftp <server IP> fdr_log

 

Core dump is easiest to retrieve from the web interface, under Troubleshooting - Core Dump.


HTH,

Arimo
HPE Networking Engineer