We have had memory leak issues for many, many years now. The platform is ASR1001HX. Back when we were running 17.3.3 TAC stated that indeed we had memory issues and to upgrade to 17.3.5. That didn't solve the problems. Over the last few years I've been watching releases and seeing what the feedback is. I've decided along the way to not upgrade and just deal with the problem we already know. Every 4-5 months a reload solves the problem.
Well now we are dealing with EOL issues and some security concerns that tip the scales in favor of upgrading. Last week I gave 17.9.5a a chance and that did not go well. Committed memory off the charts. Down to less than 1% free memory after reload. Very laggy CLI. I scrambled and read through the release notes for 17.12.4 and gave that a try. That resulted in more or less the same situation, minus the laggy CLI.
Things seem to be running okay, although memory usage is creeping the wrong way still. However, I'm concerned about the committed memory statistics. Is this working as intended, is this a cosmetic error, or do I have issues?
.hx1#sho plat soft status con bri
Load Average
Slot Status 1-Min 5-Min 15-Min
RP0 Healthy 0.54 0.30 0.20
Memory (kB)
Slot Status Total Used (Pct) Free (Pct) Committed (Pct)
RP0 Warning 7930304 7179288 (91%) 751016 ( 9%) 8926396 (113%)
One extra question, for those that are running full BGP tables (both IPv4 and IPv6) from multiple peers are you doing fine with 8G DRAM?
Thanks in advance.
