Safe harbour version 16.3.6 with a serious memoriy leak.

[rune.jon]

Sometimes picking a Cisco software version is like choosing between plague or colera without any remedy available. After the "Smart Install" vulnerability the safe harbour version for a 3650/3850 plattform went from 16.3.5b to 16.3.6. Then it was dicovered that both versions have a severe memory leak. CSCvi38191. But it was fixed in the version 16.3.6 (6.11). Is that 6.11 a public available version?

Will there be a 16.3.7 soon?

Thanks

[Leo Laohoo]

---

@rune.jon wrote:

But it was fixed in the version 16.3.6 (6.11). Is that 6.11 a public available version? 

---

No, it's an Engineering release.  TAC should be able to publish this.

[jmandersson]

I find it very intressting that the problem description says " non service impacting". Well, it is. You have three choices and all of them are certainly "service impacting":

1: Do nothing and wait for the crash.

2: Periodic reload

3: Change release

[jmhouse96]

What is the impact that you are seeing?

[Noclss2000]

I feel your pain. our standard code when putting in 3850s was 16.3.5b. Everything had been good the past ~10 months until the past week. We had 2 switch stacks hit this bug in a hospital last week. I got an on-call page on 1/16 at 3am about another switch being down, had to drive down to fix. It's always been switch 1 in the stack that loses it's mind and stops passing traffic. Pulling the power on the switch is the only way to recover. fun for a hospital environment.

I've got a surgery center now that's complaining of slowness and sure enough, switches are complaining of high memory usage. Upgrading them tonight to 16.3.7.

[jmhouse96]

I had upgraded about 40 stacks to 16.3.7 and continued to see issues with that code, my issues are around PoE and phones and APs not being able to pull PoE which normally begins to show up around 2-3 months after the upgrade. I have been running 16.6.4a for about 2 months on 8 stacks without issues yet. I am going to move forward with this code version and deploy to the rest of my enterprise in the coming weeks. 

[lem9064]

Hi,

 Wondering how things have been with 16.6.4. Our enteprisewas starting to upgrade our switches to 16.3.7 but I saw your post and didn't want to upgrade over 200 switch stacks and run into a problem. Any info you can give me would be greatly appreciated.

Thank you,

[Noclss2000]

I didn't go to 16.6.4.
The issue was in 16.3.5b and 16.3.6. The memory leak issues as well as Poe bugs were fixed in 16.3.7. So far so good in the hospital I upgraded.

[jmhouse96]

Well so far so good, we have taken more time than I would have liked to deploy across the enterprise. I have had 10-12 switch stacks running this code since about a week after the original post without issues. 

[jmhouse96]

I am now seeing some weird issues and I have now seen the Imax POE error on 16.6.5 and am awaiting further troubleshooting from TAC.

[jmhouse96]

So here is an update to this, we had updated out wireless controller to 8.8.110 and then went back to 8.8.100 as well as upgraded to 16.6.5 and this is when we had some weirdness. A few weeks later I upgraded the controllers to 8.8.120 and for the most part things settled down for the APs and phones on switches running 16.6.5, but the TAC engineer that I have been working with sent an email stating she worked with another customer seeing similar issues and they upgraded to 16.9.3 and things seemed to be resolved. 

I was on Cisco download site yesterday and noticed in order to download 16.6.5 there is now an alert that comes up saying you are aware of some serious bugs in that code. I now want to upgrade but am unsure if I should move to what TAC advised or just move to 16.6.6.

Has anyone used 16.9.3 that can provide any feedback?

[Leo Laohoo]

Thanks for the feedback. 16.6.5 is still deemed by Cisco as a "STAR" release.