Tor 0.3.2.9 Linux - first period fine, since today lots of: Your computer is too slow to handle this many circuit creation requests!
Hi All,
Is this a "known" issue, my non-exit relay has been running for over a year, and although with the recent issues (attack / network issue or the likes) with some ipfilter kunfu it managed to get through the storm pretty well. Now all of a sudden since early today my logs are flooded with:
Jan 18 11:34:44 tornode Tor[65839]: Your computer is too slow to handle this many circuit creation requests! Please consider using the MaxAdvertisedBandwidth config option or choosing a more restricted exit policy. [204844 similar message(s) suppressed in last 60 seconds]
First message is at Jan 18 07:17:13, last just Jan 18 11:37:44, when adding the # of circuits up, total in ~4 hours: 18033820 being 18 Million, which feels a bit much, but don't have any figures to compare. On the current hardware this never seemed to be an issue, not even by far it was normally (according to ESXi) at ~400-500mhz, now it's using 2 vCPU's at ~4000Mhz.
Now of course I can lower the bandwidth, but if these are low bandwidth circuits then lowering the bandwidth from 10Mbyte/sec (peak 12,5Mbyte/sec) to a lower value might actually not resolve it.
Anybody seeing this as well and/or ideas / advice etc?
Stijn
I am seeing the same issue, also for the first time, despite having run a relay for over three years. I'm willing to adjust the MaxAdvertisedBandwidth, but I'm waiting to see if this is bug that will be addressed first.
On Thu, Jan 18, 2018 at 3:44 AM, Stijn Jonker sjcjonker@sjc.nl wrote:
Hi All,
Is this a "known" issue, my non-exit relay has been running for over a year, and although with the recent issues (attack / network issue or the likes) with some ipfilter kunfu it managed to get through the storm pretty well. Now all of a sudden since early today my logs are flooded with:
Jan 18 11:34:44 tornode Tor[65839]: Your computer is too slow to handle this many circuit creation requests! Please consider using the MaxAdvertisedBandwidth config option or choosing a more restricted exit policy. [204844 similar message(s) suppressed in last 60 seconds]
First message is at Jan 18 07:17:13, last just Jan 18 11:37:44, when adding the # of circuits up, total in ~4 hours: 18033820 being 18 Million, which feels a bit much, but don't have any figures to compare. On the current hardware this never seemed to be an issue, not even by far it was normally (according to ESXi) at ~400-500mhz, now it's using 2 vCPU's at ~4000Mhz.
Now of course I can lower the bandwidth, but if these are low bandwidth circuits then lowering the bandwidth from 10Mbyte/sec (peak 12,5Mbyte/sec) to a lower value might actually not resolve it.
Anybody seeing this as well and/or ideas / advice etc?
Stijn
tor-relays mailing list tor-relays@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays
Hi everybody
Am 18-Jan-18 um 11:44 schrieb Stijn Jonker:
First message is at Jan 18 07:17:13, last just Jan 18 11:37:44, when adding the # of circuits up, total in ~4 hours: 18033820 being 18 Million
The same here: 7993419 circuits and 64009930 NTor in 4 hours (Freebsd, Jan 9th, Tor 0.3.2.8-rc)
and there without the 'too slow to handle' warning: 17059168 circuits and 37961895 NTor in 3 hours (Freebsd, Jan 14th, Tor 0.3.2.8-rc) Interesting here is the memory went up to 15GB where MaxMemInQueues was set to 2GB.
On 19 Jan 2018, at 04:59, Felix zwiebel@quantentunnel.de wrote:
Hi everybody
Am 18-Jan-18 um 11:44 schrieb Stijn Jonker: First message is at Jan 18 07:17:13, last just Jan 18 11:37:44, when adding the # of circuits up, total in ~4 hours: 18033820 being 18 Million
The same here: 7993419 circuits and 64009930 NTor in 4 hours (Freebsd, Jan 9th, Tor 0.3.2.8-rc)
and there without the 'too slow to handle' warning: 17059168 circuits and 37961895 NTor in 3 hours (Freebsd, Jan 14th, Tor 0.3.2.8-rc) Interesting here is the memory went up to 15GB where MaxMemInQueues was set to 2GB.
A few people have reported increased RAM usage recently, on and off list. It may be the next stage of the extra load. It even happens on relays with the destroy cell fix.
On my machine, Tor used too much RAM and started swapping. Please use the advice in previous threads to reduce connections and RAM.
If it keeps going, we will analyse the traffic and make code changes.
T
tor-relays@lists.torproject.org
-
Felix -
Marcel Krzystek -
Stijn Jonker -
teor