Hello Debian: This email is reporting a problem which may affect Debian distros, which report is in accordance with the directions on:
https://www.debian.org/contact.en.html#installuse Does Main-Stream Debian even use Pipewire and Wireplumber? If not, then this email might not interest you. I observed what IS a problem on various Raspberry Pi 5, 8GB (default memory / swap configuration) Running the newest Raspberry Pi OS (Debian GNU/Linux 12 -- Bookworm) at basically Standard temperatures and Pressures in clean, a low to no RF environment. I don't know is these add-on programs are germane to Raspberry Pi OS or if they might affect main-stream Debian distro users as well. If you feel I should just take this up with the folks at Raspberry Pi, they've advised me previously that the Raspberry Pi is marketed for children to teach them programming skills. I did walk away from that message feeling rather stupid that I had even been using a child's computer as a workstation. I won't bother them again... or buy any more of their product. I only utilized Pi's because my Small Home / Office (SoHo) had got broken into and someone had their way with a very expensive computer. A Raspberry Pi's media can fit in your wallet and the bad guys (burglar) can take the Pi with them: I have more of them. The problem I observed was running a lone simple BASH shell script using IPSet, i.e.: sudo ipset add MySet aaa.bbb.ccc.ddd/xx -exist That works fine for about the first 200 or so IPv4. Then it starts losing data. Wireplumber (according to Gnome System Monitor and HTOP) ends up taking more CPU time than the BASH script, where the BASH script is the only user-initiated "program" in a desktop environment. I think the problem there will be immediately obvious to you. I was adding thousands of IP's when I measured the error rate for a proof - of - concept error reoort but I had begun to see the problem, in normal usage, at about 104 IPv4, in exactly the conditions noted above. To test my theory about Wireplumber and Pipewire, I inserted "sleep 5" statements for every 10 lines of code, which code added IPv4's to MyList. The "sleep" statements were to allow this backgroud Wireplumber process to catch up, assuming linear processing of the BASH script code (which may not be the case). [code] . . [code] sleep 5 [code] . . ...etc. That extended how many sucessive (sorted) IPv4's I could add to the MyList, without problems, to about 197 IPv4's. At thousands of IPv4 (I tested above 10,000) 57.76% of the original list of numbers never made it though the IPSet statements (in a BASH script) into MyList, using only 1.7 GB of 8.2 GB. The pattern was to drop entire ranges of IPv4's and then start adding IPv4's to MyList again, further down the sorted list. I recomnend using thousands of valid IPv4, in valid CIDR notation, to ensure a clear and extended view of the problem. It's pretty simple to reproduce. (I have checked the validity of the IPv4's CIDR many times.) While the script only used about 20% of the 4 CPU's, when Wireplumber kicked in, the CPU's started running at 80% to 100%, until (presumably) the BASH script hit "sleep 5 statements. Then the load might subside briefly. In truth, the "sleep" statements may have only compounded the problem, causing the BASH script to be run in fragments while the "sleep" statements caused more of a load when they expired. The "sleep" statements are not necessary to rephoducing the problem. They were only a diagnostic step. Anyway, I'm NOT looking for tech support to solve any problem for me. I could solve the problem using IP Range (or some other means) to concatenate adjacent IPv4 ranges, thus reducing their number, or simply segmenting the IPv4's into discrete BASH scripts. That's not the point. The point is to inspire further testing BEFORE you consider incorporating Pipewire and Wireplumber into a mainstream Debian distro, if you haven't already. It's what I can do to give back to the "Giants" whom have given us Linux. I think the public by-and-large trusts that computers make astronomically few mistakes, so they're definitely not expecting errors on the order of 57% data being dropped, especially as it might pertain to an SBC being used with regard to a firewall application. How I discovered this was a firewall failed on FRIDAY 13th. ...Took out the Raspberry Pi and the server I was working on. If you'd prefer not to hear from me in the future, just send me a terse little condescendng message and I'll take the hint. Kind regards, JOHN [email protected] Sent from [Proton Mail](https://proton.me/mail/home) for Android.

