Update from 2.4.3-p1 on SG-3100 to RC-2.4.4.r.20180917.0837 crash report on startup

[luckman212]

After updating an SG-3100 from 2.4.3-p1 to the RC snapshow, I got a "Crash Report" warning when logging back in after the update was compete. Looks like just a little false positive due to maybe some extra logging that's taking place.

[occamsrazor]

[removed as realised my crash not relevant to yours]

[stephenw10]

We've opened a bug report for this:
https://redmine.pfsense.org/issues/8915

If you manage to reproduce it please drop ls -l /var/crash/ onto the bug.

Steve

[Tktowman]

I have an identical crash from 2.4.3-p1 to 2.4.4 on an SG-2440.

Output of ls -l /var/crash/ :

total 8
-rw-r--r-- 1 root wheel 5 Sep 20 10:01 minfree

[luckman212]

I looked at the code in index.php as well as crash_reporter.php and ran some tests manually from the console. Seeing your screenshot I thought it was somehow related to having the minfree file at /var/crash, but I can't replicate it.

I see a Useless Use of Cat at index.php@L64 ... could probably do away with the whole cat | wc | awk and replace it with a simple if filesize(foo) > 0 check.

And I'll admit, I don't understand why the $skip_files array (@L71) contains . and .. and a "" (null?)— what purpose does that serve? The glob("/var/crash/*") doesn't seem to add them to the array so I don't see how they would ever get hit.

But aside from little cleanups like that, this one is still a head scratcher for me.

[jimp]

@luckman212 said in Update from 2.4.3-p1 on SG-3100 to RC-2.4.4.r.20180917.0837 crash report on startup:

I looked at the code in index.php as well as crash_reporter.php and ran some tests manually from the console. Seeing your screenshot I thought it was somehow related to having the minfree file at /var/crash, but I can't replicate it.

I see a Useless Use of Cat at index.php@L64 ... could probably do away with the whole cat | wc | awk and replace it with a simple if filesize(foo) > 0 check.

And I'll admit, I don't understand why the $skip_files array (@L71) contains . and .. and a "" (null?)— what purpose does that serve? The glob("/var/crash/*") doesn't seem to add them to the array so I don't see how they would ever get hit.

But aside from little cleanups like that, this one is still a head scratcher for me.

It's on my todo list to make some changes to align how both of those pages check.

It's also entirely possible that the report you see at first is gone by the time you reach the crash reporter page, so refreshing the dashboard may no longer show the report. There is some cleanup of PHP errors done by the update process that might happen later in the boot than when the login screen becomes available. So that could be timing. Unless the dashboard keeps saying it's there repeatedly.