flock() error after upgrading

stephenw10

None of those cronjobs look unusual. I those on test boxes here and none of them show those error logs.

Maff

well bugger, just got a chance to mess around so started out with the factory reset this time configured my wan lan and error has returned

is there anything left behind in a factory reset?, hoping not to have to do a fresh install

stephenw10

When you reset the config to factory defaults it's exactly that; the config file is replaced by the default config. So, yes, potentially there could be many things left behind.

If you run at the command line: top -HaSP can you see what's being run once a minute to trigger this?

Maff

@stephenw10

just for testing i have concluded nothing i have configured or installed is causing this

just gone through a fresh hard drive in my hardware, pfsense 2.6ce installed, upgraded to 22.01, upgrade to 22.05 upgrade to 23.01rc and error is there

another fresh hard drive, 2.6ce installed, upgrade to 22.01 then straight to 23.01rc and the error is there

Only things configured was wan connection and basic default lan

Maff

Am also kinda a novice when it comes to Linux etc, alway trying to learn, I am not sure what I am looking for in the output of that command

Was a lot of things flashing up constantly but didn’t see any refresh to anything in the error output

stephenw10

Ah, interesting. So this appears to be something hardware or network related.

What are you running this on? How is the network setup that it's connected to?

Maff

@stephenw10

Running on a Lenovo m720q with a intel 9500t cpu, 16 gig of ram, network card is intel x710-t2l

Port one is dhcp setup to cable modem in modem mode 2.5gbs link

Port 2 is the basic lan setup you go through when installing pfsense 10gb link

Not noticed any issue with this hardware on 22.05

Maybe some form of sensor not working right in the new kernel?

SteveITS

@maff said in flock() error after upgrading:

Linux

FreeBSD in this case
top shows running processes.

Question: does it happen if you're NOT displaying the dashboard? Logged out of the web GUI?

Is the config file on disk growing in size or changing time stamps? (/conf/config.xml)

stephenw10

Mmm, that's a good question. Something that could be configured to update once a minute like that might be a dashboard widget.

jimp

Does anyone seeing this have some kind of off-firewall monitoring system that probes the firewall once per minute? Maybe some kind of third party integration / tacked on "API" that is leveraging xmlrpc.php?

Nothing in the base system queries that file except for HA sync, and HA sync would not be triggered that often.

blathierry

@jimp I use the home assistant integration and uptime kuma. Is that causing it?

stephenw10

How is that querying? SNMP? Some plugin code?

stephenw10

Yeah, it almost certainly is that. See: https://forum.netgate.com/post/1081590

blathierry

@stephenw10 https://github.com/travisghansen/hass-pfsense

jimp

Drop to a shell and run tail -F /var/log/nginx.log and watch a bit and see what is hitting xmlrpc.php and trace back from there. You could also grep through the older log entries to see.

You can also look at / search the log from the GUI (Status > System Logs, System tab, GUI Service sub-tab), but doing so can be tricky since accessing the GUI will add more entries to the log potentially pushing the older entries off.

renegade

@blathierry said in flock() error after upgrading:

@jimp I use the home assistant integration and uptime kuma. Is that causing it?

Interesting! I use the HA Integration also!
But from my understanding the integration polls the status every 30 seconds. The error seems to occur more often than every 30 seconds.

stephenw10

I note that code saw updates as recently as today. Are you running the current version?

jimp

Anyone hitting this, try installing the System Patches package and then create an entry for 053f60e56d09ba711af245c7f0ce6a06673ccc89 to apply the fix.

Or apply this diff/change by hand:

diff --git a/src/etc/inc/util.inc b/src/etc/inc/util.inc
index f838f6b540..2636fb4aec 100644
--- a/src/etc/inc/util.inc
+++ b/src/etc/inc/util.inc
@@ -169,12 +169,13 @@ function try_lock($lock, $timeout = 5) {
 }
 
 /* unlock configuration file */
-function unlock($cfglckkey = 0) {
-	if (!is_null($cfglckkey)) {
-		@flock($cfglckkey, LOCK_UN);
-		@fclose($cfglckkey);
-	}
-	return;
+function unlock($cfglckkey = 0)
+{
+	if (!is_resource($cfglckkey))
+		return;
+
+	flock($cfglckkey, LOCK_UN);
+	fclose($cfglckkey);
 }
 
 /* unlock forcefully configuration file */

Maff

@jimp

Thanks for looking further into this, needed a break from messing around, will test this later and I completely forgot about the home assistant project I was looking into, will test this soon as I can

blathierry

@stephenw10 Yeah, newest version still does it sadly. I updated both of them. I disabled the home assistant integration for now.. I hope the maintainer will fix it soon, it's very unexpected.