SABnzbd+ hangs after episode of CPU starvation
Posted: September 19th, 2010, 12:20 pm
Version: 0.5.2 final; 0.5.3 final; 0.5.4 final
Hardware: QNAP NAS (Arm) TS-410
OS: QNAP linux (2.6.31 variant)
Install source: NAS package (+ updated to 0.5.3 and 0.5.4 with source packages)
Firewall: No
IPv6: No
Reproducible: Yes, but after a variable length of time
Skin: Classic
Problem description
When SABnzbd+ is experiencing very slow operation due to severe system load for a period of time (> 10 minutes), SABnzbd+ will stop downloading, and become unresponsive to admin commands. Normal operation is not resumed once system load returns to normal, and SABnzbd+ must be killed with 'kill -9'.
When SAB has entered its malfunctioning state, the display will remain responsive (when CPU load has returned to normal). The last download speed will remain in the display (e.g. 650 kB/s) but will not update, and no data is downloaded. Multiple error log entries are made stating 'missing authentication' e.g.
2010-09-05 11:13:14,560::WARNING::[__init__:1309] Authentication missing, please enter username/password from Config->General into your 3rd party program:
2010-09-05 11:40:05,237::WARNING::[__init__:1309] Authentication missing, please enter username/password from Config->General into your 3rd party program:
In some cases, when I have used additional logging, I get a bunch of 'file missing from all servers' warnings, shortly before the first 'authentication' error - but I haven't confirmed that they are connected, or that the articles genuinely are missing.
Admin commands (e.g. restart/shutdown) do not work, and appear to have no effect.
The problem is made dramatically worse if SAB is run 'nice'. E.g. running with nice 20, seems to provoke it, and running with higher priority mitigates it (but a severe cost to other more interactive processes)
My guess is that a timeout somewhere is getting triggered, and not properly handled - but I don't know how to debug this further.
Hardware: QNAP NAS (Arm) TS-410
OS: QNAP linux (2.6.31 variant)
Install source: NAS package (+ updated to 0.5.3 and 0.5.4 with source packages)
Firewall: No
IPv6: No
Reproducible: Yes, but after a variable length of time
Skin: Classic
Problem description
When SABnzbd+ is experiencing very slow operation due to severe system load for a period of time (> 10 minutes), SABnzbd+ will stop downloading, and become unresponsive to admin commands. Normal operation is not resumed once system load returns to normal, and SABnzbd+ must be killed with 'kill -9'.
When SAB has entered its malfunctioning state, the display will remain responsive (when CPU load has returned to normal). The last download speed will remain in the display (e.g. 650 kB/s) but will not update, and no data is downloaded. Multiple error log entries are made stating 'missing authentication' e.g.
2010-09-05 11:13:14,560::WARNING::[__init__:1309] Authentication missing, please enter username/password from Config->General into your 3rd party program:
2010-09-05 11:40:05,237::WARNING::[__init__:1309] Authentication missing, please enter username/password from Config->General into your 3rd party program:
In some cases, when I have used additional logging, I get a bunch of 'file missing from all servers' warnings, shortly before the first 'authentication' error - but I haven't confirmed that they are connected, or that the articles genuinely are missing.
Admin commands (e.g. restart/shutdown) do not work, and appear to have no effect.
The problem is made dramatically worse if SAB is run 'nice'. E.g. running with nice 20, seems to provoke it, and running with higher priority mitigates it (but a severe cost to other more interactive processes)
My guess is that a timeout somewhere is getting triggered, and not properly handled - but I don't know how to debug this further.