Hi, I'm running FreeRADIUS 2.1.11 on Gentoo compiled with specific patches (qafixes, versionless, pkglibdir) and a small patch of my own (byminute, adds an extra var in xlat.c, nothing big). I'm experiencing a weird crash of which I've found absolutely nothing online: "Failed creating handler". Source code says this one comes from src/main/event.c when calling fr_event_insert() but I can't figure out anything else. The server does decoupled accounting, one site has only one module in accounting, rlm_detail and the other listens on the detail logs with only one module in accounting, rlm_python. cleanup_delay is 5, max_requests 10240, 16 threads, max_requests_per_server = 1500 Any idea what could be the problem? tx, amne
Hi,
I'm running FreeRADIUS 2.1.11 on Gentoo compiled with specific patches (qafixes, versionless, pkglibdir) and a small patch of my own (byminute, adds an extra var in xlat.c, nothing big). I'm experiencing a weird crash of which I've found absolutely nothing online: "Failed creating handler". Source code says this one comes from src/main/event.c when calling fr_event_insert() but I can't figure out anything else.
My new 2.1.11 died after about 24h of happy RADIUSing - twice now. It's too busy to run -X, so I don't have a lot of logs. radius.log logs the last previous auth OK - and then the process is gone. Would this behaviour fit to this problem cause? Worth trying the usec fix in GIT? Greetings, Stefan Winter
The server does decoupled accounting, one site has only one module in accounting, rlm_detail and the other listens on the detail logs with only one module in accounting, rlm_python. cleanup_delay is 5, max_requests 10240, 16 threads, max_requests_per_server = 1500
Any idea what could be the problem?
tx, amne - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
On Sat, Jun 25, 2011 at 5:04 PM, Alan DeKok <aland@deployingradius.com> wrote:
Stefan Winter wrote:
Would this behaviour fit to this problem cause?
Worth trying the usec fix in GIT?
The fix for rlm_detail <sigh>
Alan, would it be better if you release 2.1.11.1 (or something like that) with that fix (commit 731e733) included? -- Fajar
Alan DeKok <aland@deployingradius.com> wrote:
Stefan Winter wrote:
Would this behaviour fit to this problem cause?
Worth trying the usec fix in GIT?
The fix for rlm_detail <sigh>
Just so you know, that was the bug causing my poor ARM boxen to OOM kill FreeRADIUS. At least I now do not need to install an experimental armel valgrind :) Cheers -- Alexander Clouter .sigmonster says: Expect the worst, it's the least you can do.
participants (5)
-
Alan DeKok -
Alexander Clouter -
Cornel Cruceru -
Fajar A. Nugraha -
Stefan Winter