@HPoltergeist
Are you running on HDD?
Have you tuned the network on your host?
Config LXC 1024 GB 8GB some tuning on host was created with gui
https://tricorder.pi-hole.net/BFNfNiiP/
My dns blaster justs asks for google.com and steps thru the services…
---- FINAL STATS ----
Sent: 19
Received: 19
RecvErr: 0
WriteErr: 0
BusyID: 0
UnknownResp: 0
ShortResp: 0
Elapsed: 20.30s
Achieved send QPS: 0.9
Reply ratio: 100.00%
Latency samples: 19
p50 latency: 0.88 ms
p95 latency: 66.85 ms
p99 latency: 66.85 ms
LXC 512 GB 8 GB some tuning on host was Created with gui
https://tricorder.pi-hole.net/3BigQPfM/
---- FINAL STATS ----
Sent: 19
Received: 19
RecvErr: 0
WriteErr: 0
BusyID: 0
UnknownResp: 0
ShortResp: 0
Elapsed: 20.30s
Achieved send QPS: 0.9
Reply ratio: 100.00%
Latency samples: 19
p50 latency: 3.97 ms
p95 latency: 67.23 ms
p99 latency: 67.23 ms
curl failed uploading report But works ok with good stats was manually created.
pct create xxx local:vztmpl/debian-12-standard_12.12-1_amd64.tar.zst \
--hostname piholexx \
--cores 2 \
--memory 8192 \
--rootfs local-lvm:8 \
--net0 name=eth0,bridge=vmbr0,ip=192.168.1.90/24,gw=192.168.1.1 \
--password yourpassword \
--unprivileged 1
pct start xxx
I had to add the gateway afterwards in the lxc.
I had to set ip and gw on host for curl to work
https://tricorder.pi-hole.net/8TPjO9vY/
---- FINAL STATS ----
Sent: 19
Received: 19
RecvErr: 0
WriteErr: 0
BusyID: 0
UnknownResp: 0
ShortResp: 0
Elapsed: 20.30s
Achieved send QPS: 0.9
Reply ratio: 100.00%
Latency samples: 19
p50 latency: 1.85 ms
p95 latency: 3.93 ms
p99 latency: 3.93 ms
RE: 429's
Thanks for saying something, I assumed I was the only one seeing them because I only get them on the admin side of the application. I'll take a look at the rate limiting mixin and see why it has suddenly become more sensitive.
smokingwheels:
The host system is running on NVMe SSDs.
No special networking settings on the host or CT.
All other CT/VM goes without issues.
Local network is full 1000Mbps, internet connection is 500/50.
DanSchaper:
No worries, happy to help! Also thanks for looking into it!
@HPoltergeist
Do the pihole DNS queries go much past 10k to 15k per 10 min?
That’s all I can get out mine before the CT CPU stall starts to spike red.
DNS queries
No, not past 10k /10 min, way lower.
The typical values are between 200-400 /10 min, with busier periods of 500-700 /10 min.
Some spikes up until 900 /10 min, but not exceeding 1000.
Around a total queries of 50k /24h, for 10-16 clients connected at a time.
As mentioned, I am not experiencing hardware/resource exhaustion or bottlenecks, the CPU usage (2 cores) on the CT is around 0,1-0,5% at all times.
All right, so I would like to thank you guys for all the help and troubleshooting!
So far it seems that I don’t experience the mentioned issues anymore.
The most probable factor in resolving this so far, seems to be the switch to Ubuntu instead of Debian.
(Flagging this as solution for now.)
I hope this thread will be helpful in a way or another for others wandering around here.
This way, I believe we can let this topic get closed.
Thank you once again!
In the meantime we have noticed another hiccup what could be related to this original issue.
Currently, we have the DHCP lease set to 24h with MAC-IP binding with additional DNSMASQ names.
Multiple clients lost network connectivity independently at different times.
When we started investigating, the only connection between the cases is that it happens at the time (respectively) when the client gets a new DHCP lease.
At these times usually a NIC restart or reconnecting to Wifi solves the issue on the client.
This affects multiple devices, so far:
- mobile Wifi x2
- laptop Wifi x1
- laptop Eth x1
At the moment it is not the end of the world, but I believe this would worth another ticket to have this investigated on your (dev) side too.