Posted on September 25, 2020 6:00 am
 |  Asked by Martin Wolf
 |  39 views
0
0
Print Friendly, PDF & Email

Hello,

Recently we discovered plenty of zipped Sflow.gz and netns.gz files at var/core on the Arista switch.

And when logging on we get this Warning:

 

Warning: the following filesystems have less than 10% free space left:
tmpfs (on /var/core) 0% (0 Available)
Please remove configuration such as tracing and clean up the space.

 

 

#bash df -h | grep var/core
tmpfs 792M 792M 0 100% /var/core

Please advise.

 

0
Answered on September 25, 2020 6:10 am

Hi Martin,

It looks like the sFlow agent is crashing and creating core dump files. If the /var/core is full, there is a good possibility of continuous repeated crashes. There can be multiple reasons for the crash like memory leak, bugs, unsupported config etc and we need to analyze some logs before concluding on the reason for the crash.

Could you please open a case with TAC either by sending an email to support@arista.com or via the customer portal with details.

Thanks,
Bharath

0
Posted by Martin Wolf
Answered on October 23, 2020 5:28 pm

It seems to hit the following bug:

 

Bug 427674:
When sampling packets with the BGP extension enabled, if the address family of the route does not match the address family of the next hop, the sFlow agent can restart unexpectedly. Use "no sflow extension bgp" to stop the sFlow agent agent from restarting.

 

Affected releases
4.23 - 4.23.2 to 4.23.5
4.24 - 4.24.0 to 4.24.2

 

Workaround:

When sampling packets with the BGP extension enabled, if the address family of the route does not match the address family of the next hop, the sFlow agent can restart unexpectedly. Use "no sflow extension bgp" to stop the sFlow agent agent from restarting.

Post your Answer

You must be logged in to post an answer.