hi, I did send a message to this list earlier with more information about where our 3.x install is seg faulting....but the list doesnt like bigger than 100k attachments - the log from gdb being around 260k . shall I send to another place? cheers alan
On 5 Dec 2012, at 23:11, Alan Buxey <A.L.M.Buxey@LBORO.AC.UK> wrote:
hi,
I did send a message to this list earlier with more information about where our 3.x install is seg faulting....but the list doesnt like bigger than 100k attachments - the log from gdb being around 260k . shall I send to another place?
http://bugs.freeradius.org Probably stick it in a gist, or add it in a ``` ``` block. Are these new segfaults or old segfaults? Because after the proxy rewrites AlanD did I don't see the server down segfaults anymore... -Arran
On 5 Dec 2012, at 23:39, Arran Cudbard-Bell <a.cudbardb@freeradius.org> wrote:
On 5 Dec 2012, at 23:11, Alan Buxey <A.L.M.Buxey@LBORO.AC.UK> wrote:
hi,
I did send a message to this list earlier with more information about where our 3.x install is seg faulting....but the list doesnt like bigger than 100k attachments - the log from gdb being around 260k . shall I send to another place?
Probably stick it in a gist, or add it in a
```
```
block.
Are these new segfaults or old segfaults? Because after the proxy rewrites AlanD did I don't see the server down segfaults anymore...
In fact after a few fixes 3.0.0 with the new rlm_sql and rlm_ldap appears pretty stable. I have seen a few weird lockups (temporary) and issues with EAP session matching, which is why I just added a bunch more debugging statements to mem.c to try and figure out what's happening. Keep seeing random blocks of: Wed Dec 5 13:27:12 2012 : Error: rlm_eap: No EAP session matching the State variable. Wed Dec 5 13:27:12 2012 : Error: rlm_eap: No EAP session matching the State variable. Wed Dec 5 13:27:17 2012 : Error: rlm_eap: No EAP session matching the State variable. Wed Dec 5 13:27:17 2012 : Error: rlm_eap: No EAP session matching the State variable. Wed Dec 5 13:27:22 2012 : Error: rlm_eap: No EAP session matching the State variable. Wed Dec 5 13:27:22 2012 : Error: rlm_eap: No EAP session matching the State variable. Wed Dec 5 13:27:24 2012 : Error: (33531) Received conflicting packet from client x.x.x.x/16 port 32770 - ID: 235 due to unfinished request. Giving up on old request. Wed Dec 5 13:27:24 2012 : Error: rlm_eap: No EAP session matching the State variable. Wed Dec 5 13:27:24 2012 : Error: rlm_eap: No EAP session matching the State variable. Wed Dec 5 13:27:27 2012 : Error: rlm_eap: No EAP session matching the State variable. Wed Dec 5 13:27:27 2012 : Error: rlm_eap: No EAP session matching the State variable. I'll be interested to hear your experiences. -Arran
Did that locking code go public? Arg... It was supposed to be for internal testing of a feature in develeopment. Sent from my iPhone On 2012-12-06, at 5:18 AM, Phil Mayers <p.mayers@imperial.ac.uk> wrote:
On 12/05/2012 11:55 PM, Arran Cudbard-Bell wrote:
Keep seeing random blocks of:
Hmm. I notice Alan added locking to the core rbtree code. Did anyone *remove* the duplicate locking in mem.c following that? - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/devel.html
On 6 Dec 2012, at 12:53, Alan DeKok <aland@DEPLOYINGRADIUS.COM> wrote:
Did that locking code go public? Arg... It was supposed to be for internal testing of a feature in develeopment.
Yeah. https://github.com/FreeRADIUS/freeradius-server/commit/42c837c224384cd3aff1f... Wasn't me :P -Arran
On 6 Dec 2012, at 10:18, Phil Mayers <p.mayers@IMPERIAL.AC.UK> wrote:
On 12/05/2012 11:55 PM, Arran Cudbard-Bell wrote:
Keep seeing random blocks of:
Hmm. I notice Alan added locking to the core rbtree code. Did anyone *remove* the duplicate locking in mem.c following that?
No, not yet, but the locking isn't applied by default. -Arran
On 6 Dec 2012, at 13:04, Alan Buxey <A.L.M.Buxey@LBORO.AC.UK> wrote:
Hi,
segfault in 3.x as previously discussed:
Could you try with current HEAD? -Arran
Hi,
okay - the original email has now been dropped/blocked due to the 100k limit. will look at uploading it elsewhere for analysis
Are these new segfaults or old segfaults? Because after the proxy rewrites AlanD did I don't see the server down segfaults anymore...
old problems....our 3.x box dies all the time - we have a keepalive system to kick it back into action. this bug/issue might be the only thing causing the outages.... or there may be others but need to get this one out of the way first. the build is from the code a couple of weeks ago (when we had that rlm_perl discussion). alan
participants (4)
-
Alan Buxey -
Alan DeKok -
Arran Cudbard-Bell -
Phil Mayers