Posted on September 28, 2020 7:49 pm
 |  Asked by Kevin Malone
Print Friendly, PDF & Email

I have a factory reset arista 7010t I wanted to use for a project runing version 4.23.0F.  Upon bootup i’m receiving the error message “Sep 28 17:47:02 localhost StrataCentral: %CAPACITY-1-UTILIZATION_HIGH: IFP-Slice-1-Linecard0/0 table utilization is currently at 94%, crossed threshold 90%”.

My question is why is my utilization full when this is a factory reset and fresh switch?  It’s litterally doing nothing and something is already full?  I have attached what I beleive to be relivant info.  Please help?



Posted by Sreelekha
Answered on September 29, 2020 4:07 am

Hi Kevin,

Thanks for writing to us.

Due to a cosmetic bug we see this log on 7010 and few other models when they come up after a reload/upgrade. This is due to the transient state of EOS during the initial boot up.

Bug details:

This is purely cosmetic if you see this only during the boot-up, unless you see the log repeating even after the switch is up for a few hours.


Answered on September 29, 2020 5:17 am

Hi Kevin,

Thanks for reaching out.

IFP Slice-0 and Slice-1 are utilized by the system.

IFP is used for Ingress IPV4/V6 ACLs, PBRs, etc and this model(7010T) has 16 slices capable of 256 entries each.

Once the system has detected that a slice is full, it will automatically select a new available slice to use next. Currently, your device is reporting that slice-1 is 94% full. Since we still have 15 other slices that are not full and overall IFP usage is only around 10% these logs is not a concern and we can ignore the above log msgs safely.

However if there are completely no usable Slices that PBR/ACL's/other features can use to program additional entries, then those entries would remain unprogrammed due to TCAM slice exhaustion.



Posted by Kevin Malone
Answered on September 30, 2020 6:53 pm

I understand, thank you.  It is a little alarming during a clean boot up to see something report as full.  I would think a warning would only be displayed once all slices are full?  Like if it reported on the overall IFP usage and not the individual slices.

Answered on October 1, 2020 2:23 am

Hi Kevin,

Thank you for your response.

This is purely cosmetic issue after the initial bootup (or) after upgrade.

By default, for TCAM tables, hardware capacity default thresholds are 90% for every slice in every stage, which causes a message to be logged whenever each slice reaches 90% utilization.

Also there is a way to disable the logging alert with the following command:

switch(conf)#hardware capacity alert table ifp feature all threshold 0 <---- Here 0 means 'disable the logging'

Incase if we issue the above command then it won't generate the logging alert even if last available slice gets fully consumed.



Posted by Kevin Malone
Answered on October 1, 2020 8:34 pm

got it, thanks for the info!

Post your Answer

You must be logged in to post an answer.