Dear Community,
today my ipfire installation with Core Update 199 crashed cause my disk ran full. It seems that my like 100mb per week /var/log/messages file was growing since 3 or 4 days to about 7.2gb and my disk was full. Yes i do not have the biggest disk but what it seems is that i had a lot of messages like this:
Jan 10 14:51:42 ipfire hostapd: nl80211: Beacon event message available
Jan 10 14:51:42 ipfire hostapd: green0p0: Event RX_MGMT (18) received
Jan 10 14:51:42 ipfire hostapd: unknown vendor specific information element ignored (vendor OUI 00:03:7f len=9)
Jan 10 14:51:42 ipfire hostapd: unknown vendor specific information element ignored (vendor OUI 8c:fd:f0 len=8)
Jan 10 14:51:42 ipfire hostapd: unknown vendor specific information element ignored (vendor OUI 8c:fd:f0 len=22)
Jan 10 14:51:42 ipfire hostapd: unknown vendor specific information element ignored (vendor OUI 00:04:0e len=12)
After that i checked via the command âhostapd_cli -i âgreen0p0â log_levelâ that the current loglevel was set to
[]# hostapd_cli -i 'green0p0' log_level
Current level: EXCESSIVE
Timestamp: 0
Which then i switched with the command to âhostapd_cli -i âgreen0p0â log_level ERRORâ to just error logs.
Now hostapd stoped writing endless times the same events and the /var/log/messages seems to grow normal again. I have no clue why the log level was set this way since i did used the web gui to config hostapd and changed my password to not include any special characters. Saw that was an issue before.
As always any hints or ideas what i could have done wrong or what could be the reasong why this message basically kept popping up would be great.
Thank you.
There was a bug in the hostapd changes in CU199. The root cause is being investigated and a solution will be identified.
Unfortunately no users that use hostapd did any evaluation during the Testing phase of CU199 so the problem was only identified when CU199 had full release.
That is interesting. The log_level set in the hostapd.conf file is set to 3 which is Notifications and that hasnât changed since at least 2013 and probably much earlier.
I have found some comments that hostapd may not always take any notice of the log level due to the implementation of log level in hostapd being incomplete.
Maybe the hostapd developers made some change that over-rides or ignores the log level set in hostapd.conf between the last version (64d60bb4) and the current update (f747ae0)
I can confirm that with Core Update 199 hostapd log has increased to 50-100 entries per second.
But I can also confirm that the interim solution form @brottus is working on my system
Do you guys have debug logging enabled by any chance?
There is a DEBUG variable in /var/ipfire/wlanap/settings. If that is set to anything larger than zero (or empty) it will increase the debug logging level.
I just checked. My settings file contained DEBUG=1. Setting it to 0 stopped the the excessive logs.
Thanks for the hint.
Remains the reason, why this setting existed.
The wiki isnât updated yet. Therefore it is possible to look easily at the fields of the CU198 hostapd. There is an entry âloglevelâ defined, which doesnât exist in CU199. Is this the reason for the issue?
Loglevel was removed from wlanap.cgi in CU188 at the same time as Debuglevel. If that entry was used previously then you could have an entry in settings of DEBUG=0 down to DEBUG=4 with 0 being verbose and 4 being warnings with the default being 4 so at the minimum anybody using wlanap prior to CU188 would likely have some DEBUG entry in settings.
Anybody starting with wlanap from CU188 onwards would not have a DEBUG entry in settings.
Anybody running with wlanap with CU187 or earlier would have at a minimum an entry of DEBUG=4 (the default) but it could be a more verbose number if that had been selected at some time.
The method in CU187 and earlier had 4 as warnings only and 0 as verbose with 4 as the default, so that would mean a default entry from CU187 or earlier would results in -d -d -d -d.
Thank you everyone for your swift support. Indeed i was using ipfire CU187 and before and i am pretty sure i set the value to warning only. I keept updating ipfire since therefor my DEBUG value was set to 4, hence the debug logs. @bonnietwin do you know if there is any other value which would need a clean up, since this is not a standard feature to enable debug through the webui anymore i will most likely have other legacy features too. Or is there a swift way to recreate the config ?
Again thanks for the quick feedback everyone.