Hello everyone, I am hosting 3 VM's limited at 10Mbps all together. Each VM is limited to 1Mbps via proxmox. I have noticed if i have these relays running it kills a 10Gbps fiber optic line. All the way down to 50Mbps or worse depending on what the time of day. Any idea what i can try? I noticed this happen over the past few months maybe its increased usage on the relays not sure. According to TOR relay search the demand has spiked recently. I wondering why/how it could bypass the limits on both proxmox and pfsense. Thanks, John Csuti (216) 633-XXXX
Hi, Have you set the 'RelayBandwidthRate' and 'RelayBandwidthBurst' configuration parameters in your torrc? If not then you should. ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ On Friday, April 9, 2021 6:25 PM, John Csuti <postmaster@coolcomputers.info> wrote:
Hello everyone,
I am hosting 3 VM's limited at 10Mbps all together. Each VM is limited to 1Mbps via proxmox. I have noticed if i have these relays running it kills a 10Gbps fiber optic line. All the way down to 50Mbps or worse depending on what the time of day. Any idea what i can try? I noticed this happen over the past few months maybe its increased usage on the relays not sure. According to TOR relay search the demand has spiked recently. I wondering why/how it could bypass the limits on both proxmox and pfsense.
Thanks, John Csuti (216) 633-XXXX
Hi,
if i have these relays running it kills a 10Gbps fiber optic line
this sounds like a configuration issue on your router, relays tend to establish tons of concurrent connections and some routers either can't handle it or think it's some sort of attack - especially if you run three relays all routed through the same router - the relays will likely end up sharing source and destination IP's, making it even more likely to trigger built-in filters. Do you have any reason to rate-limit the entire virtual machine, instead of just the tor instance? If not, I'd advise you to deploy rate limiting by using the tor configuration file, instead of Proxmox - last time I used PVE the rate-limiting was severely broken, and on some customers VM's, it even broke completely - no rate limiting being applied at all or network access being cut off randomly, despite having sane settings. Tor has plenty of variables you can tune to rate-limit each instance, take a look: https://2019.www.torproject.org/docs/tor-manual.html.en#BandwidthRate https://2019.www.torproject.org/docs/tor-manual.html.en#AccountingMax
Each VM is limited to 1Mbps via proxmox.
Assuming this was just a typo, but if not, please fix it or get rid of PVE's built in rate-limiting entirely. - William On 09/04/2021, John Csuti <postmaster@coolcomputers.info> wrote:
Hello everyone,
I am hosting 3 VM's limited at 10Mbps all together. Each VM is limited to 1Mbps via proxmox. I have noticed if i have these relays running it kills a 10Gbps fiber optic line. All the way down to 50Mbps or worse depending on what the time of day. Any idea what i can try? I noticed this happen over the past few months maybe its increased usage on the relays not sure. According to TOR relay search the demand has spiked recently. I wondering why/how it could bypass the limits on both proxmox and pfsense.
Thanks, John Csuti (216) 633-XXXX
Hello, We are trying to limit it now using the rate limiter inside torrc. And I don’t have a modem of any sorts. This runs on a data center connection. Also this is fiber optic on a 10Gb line to the internet backbone there isn’t a issue with too many connections. My edge router is only using about half the connections. Also my ISP has only used a fourth of the total connections. I am also working with the renter of the VM’s to figure it out. Thanks, John Csuti (216) 633-XXXX
On Apr 10, 2021, at 11:05 AM, William Kane <ttallink@googlemail.com> wrote:
Hi,
if i have these relays running it kills a 10Gbps fiber optic line
this sounds like a configuration issue on your router, relays tend to establish tons of concurrent connections and some routers either can't handle it or think it's some sort of attack - especially if you run three relays all routed through the same router - the relays will likely end up sharing source and destination IP's, making it even more likely to trigger built-in filters.
Do you have any reason to rate-limit the entire virtual machine, instead of just the tor instance?
If not, I'd advise you to deploy rate limiting by using the tor configuration file, instead of Proxmox - last time I used PVE the rate-limiting was severely broken, and on some customers VM's, it even broke completely - no rate limiting being applied at all or network access being cut off randomly, despite having sane settings.
Tor has plenty of variables you can tune to rate-limit each instance, take a look:
https://2019.www.torproject.org/docs/tor-manual.html.en#BandwidthRate https://2019.www.torproject.org/docs/tor-manual.html.en#AccountingMax
Each VM is limited to 1Mbps via proxmox.
Assuming this was just a typo, but if not, please fix it or get rid of PVE's built in rate-limiting entirely.
- William
On 09/04/2021, John Csuti <postmaster@coolcomputers.info> wrote: Hello everyone, I am hosting 3 VM's limited at 10Mbps all together. Each VM is limited to 1Mbps via proxmox. I have noticed if i have these relays running it kills a 10Gbps fiber optic line. All the way down to 50Mbps or worse depending on what the time of day. Any idea what i can try? I noticed this happen over the past few months maybe its increased usage on the relays not sure. According to TOR relay search the demand has spiked recently. I wondering why/how it could bypass the limits on both proxmox and pfsense. Thanks, John Csuti (216) 633-XXXX
tor-relays mailing list tor-relays@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays
Hey are you running this at home? I noticed that my ISPs DOCSIS modem handles a permanent traffic with many many connections very badly. Or are you saying they really use 50Mbps -> Set the traffic limit in your TOR configuration. Greetings, Sebastian On 09/04/2021 19:25, John Csuti wrote:
Hello everyone,
I am hosting 3 VM's limited at 10Mbps all together. Each VM is limited to 1Mbps via proxmox. I have noticed if i have these relays running it kills a 10Gbps fiber optic line. All the way down to 50Mbps or worse depending on what the time of day. Any idea what i can try? I noticed this happen over the past few months maybe its increased usage on the relays not sure. According to TOR relay search the demand has spiked recently. I wondering why/how it could bypass the limits on both proxmox and pfsense.
I am hosting 3 VM's limited at 10Mbps all together. Each VM is limited to 1Mbps via proxmox. I have noticed if i have these relays running it kills a 10Gbps fiber optic line. All the way down to 50Mbps or worse depending on what the time of day. Any idea what i can try? I noticed this happen over the past few months maybe its increased usage on the relays not sure. According to TOR relay search the demand has spiked recently. I wondering why/how it could bypass the limits on both proxmox and pfsense. I am experiencing a similar effect since a few months, often without filling even half of the available bandwidth. Like what Sebastian suggested earlier, I suspect the horrible quality ISP-supplied modem/router. It’s not like the bandwidth is actually used: in fact during the interruptions it falls to a negligible level. The situation is more about packets being dropped or experiencing extremely large processing times (hundreds to thousands msecs).
If your situation is similar and you do not need the router function, does switching to bridge mode help in any way?
Dear all helping, I believe I have found the issue after talking with one of my (least favorite) techs. He was doing a CPU swap on a rack router and got thermal paste onto the socket. Since all he had on had was a microfiber cloth and IPA. He assumed it would be fine to clean it without covering the pins if he was careful enough. The microfiber cloth caught some pins and (like velcro) bent them too far out of line for us to repair. I have a new mother board being sent up but it won’t be here until the 16-23 including a new case and power supply all coming between 16-30. He never reported it due to the system turning on and appearing to work. I believe some of the pins contribute to PCIe lanes or something of the sort. I am not 100% on this being the issue but it so seems reasonable enough to give it a shot fixing. I have noticed green artifacts on the screen when booting and the graphics is on board the CPU (Integrated intel graphics.) Thanks, John Csuti (216) 633-XXXX
On Apr 12, 2021, at 6:23 AM, mpan <tor-1qnuaylp@mpan.pl> wrote:
I am hosting 3 VM's limited at 10Mbps all together. Each VM is limited to 1Mbps via proxmox. I have noticed if i have these relays running it kills a 10Gbps fiber optic line. All the way down to 50Mbps or worse depending on what the time of day. Any idea what i can try? I noticed this happen over the past few months maybe its increased usage on the relays not sure. According to TOR relay search the demand has spiked recently. I wondering why/how it could bypass the limits on both proxmox and pfsense.
I am experiencing a similar effect since a few months, often without filling even half of the available bandwidth. Like what Sebastian suggested earlier, I suspect the horrible quality ISP-supplied modem/router. It’s not like the bandwidth is actually used: in fact during the interruptions it falls to a negligible level. The situation is more about packets being dropped or experiencing extremely large processing times (hundreds to thousands msecs).
If your situation is similar and you do not need the router function, does switching to bridge mode help in any way?
_______________________________________________ tor-relays mailing list tor-relays@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays
participants (5)
-
Arlen Yaroslav -
John Csuti -
mpan -
Sebastian Elisa Pfeifer -
William Kane