Forum Discussion

Shripaty's avatar
Shripaty
Icon for Cirrus rankCirrus
Sep 04, 2023

AS3 Best Practice

hi ,

A quick question , I am new to AS3 and I have observed that we are gradually increasing our load to the current f5 Environment with gradual increase in REST API Calls towards the device. Is there any way to measure the CPU load caused by these REST API calls. I have provisioned the Management with a large memory of 2048 Mb but still I see lots of errors when one tenant tries to fetch or implements any new changes in the architecture. f5 best practice is to increase the timeout but these best practices will end up resulting in a more CPU demand. 

Will increasing the timeout value for a REST call to 180 not affect the performance and CPU allocation of f5.

My current version of AS3 is 3.28.0

Current TMOS version is 16.1.3.2

Module LTM 

Tenant Application : Postgress

Any knowledge sharing would be appreciated.

If you find that the REST API is timing out, you can increase the timeout values for ircd, restjavad, and restnoded. These timeouts may occur due to large responses, such as when requesting the status of all virtual servers or all Wide-IPs. Not all variables are present in all BIG-IP versions. See K94602685 and the associated Bug ID 858189 for information on specific versions.

To increase timeout values, use the following commands (the defaults are all 60 seconds):

  • tmsh modify sys db icrd.timeout value 180
  • tmsh modify sys db restjavad.timeout value 180
  • tmsh modify sys db restnoded.timeout value 180

And then save changes and restart related services:

  • tmsh save sys config
  • tmsh restart sys service restjavad
  • tmsh restart sys service restnoded 
devops
  • Zdenek's avatar
    Zdenek
    Icon for Cirrostratus rankCirrostratus
    Feb 05, 2024

    Shripatymay I ask you how this case continues? We send extensive amount of API calls to our F5s, but they're all Rest API calls (imperative), and that probably does not cost so much CPU as the declarative API the AS3 uses. We're forced to switch to AS3 due to NEXT OS slowly approaching, but I wonder about the consequences and "lots of errors" is something I don't experience now with Rest API and definitely don't want to with AS3. Thank you

    • JRahm's avatar
      JRahm
      Icon for Admin rankAdmin
      Feb 06, 2024

      Hi Zdenek Just to be clear, BIG-IP Next will not have the same control plane challenges with API calls that exist with classic BIG-IP. 

      The issue at scale with both AS3 (which is an iControl LX package) and iControl REST is the shared restjavad process, so a lot of calls with heavy payloads will be impactful either way and need to be managed accordingly. If you aren't having issues today, I don't anticipate that will change with your shift to AS3, but testing in a lab on large configs will be helpful to determine that in your environment.

  • Shripaty's avatar
    Shripaty
    Icon for Cirrus rankCirrus
    Feb 13, 2024

    Hi Jason, 

     

    Thanks for replying , is there a way we can measure the amount of memory being utilized by those calls or any event to be identified if the device is running low on memory or exhaustive stages of memory. I have encountered issues where if I do a traffic failover or a Config Sync ends up making the devices in Peer Time out Sync. Not sure if this is due to memory exhaustion while the changes are being in Sync between the pairs.