Thread Info
  • State Verified Answer
  • Locked Locked
  • Replies 92 replies
  • Subscribers 8 subscribers
  • Views 23084 views
  • Users 0 members are here
Options
Suggested
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Weird UTM freezes randomly approximately once a day ...

Escape75

I have experienced a strange lockup on my "new" UTM box, but I checked log files and they don't reveal anything, just a bunch of weird characters ...

2023:03:16-01:32:01 escape75 /usr/sbin/cron[25494]: (root) CMD (  nice -n19 /usr/local/bin/gen_inline_reporting_data.plx)
2023:03:16-01:35:01 escape75 /usr/sbin/cron[25649]: (root) CMD (   /usr/local/bin/reporter/system-reporter.pl)
�����������������������������������������������������������������������������������������������������������
2023:03:16-09:03:10 escape75 syslog-ng[4942]: syslog-ng starting up; version='3.4.7'
2023:03:16-09:03:12 escape75 ddclient[5361]: WARNING:  cannot connect to checkip.dyndns.org:80 socket: IO::Socket::INET: Bad hostname 'checkip.dyndns.org'
2023:03:16-09:03:24 escape75 system: System was restarted



So,- I've been running the software version of UTM (9.714) on my old unit (an XG115 r2) for a couple of years without any issues,
and recently I have migrated my saved config over to a new unit (XG115 r3) and a few hours after setting up the new unit (at night)
it froze up, and interfaces were not pingable (LAN) so I powered it down and rebooted. It's working again ...

Just wondering if there's something more I can look at to see what the issue was .. I have a hunch maybe it was DHCP related,
as my devices on the LAN were renewing the IP addresses and they were not in the table on the new unit, but it's a wild guess,
so if this doesn't happen again then maybe it's nothing to worry about.

I don't know if there would be an issue moving the config file (and license) from the old unit, but I wouldn't think so.

The new unit was installed the same way as the old unit, using the ssi-9.714-4.1.iso file and removing the /etc/asg with a software license,
and the old unit hasn't experienced any weird issues in years, and the ethernet ports and devices are setup in an identical way, nothing changed.

Just looking for thoughts and ideas ...

Stats from top:

top - 11:32:20 up 2:31, 1 user, load average: 0.09, 0.29, 0.25
Tasks: 163 total, 1 running, 160 sleeping, 0 stopped, 2 zombie
Cpu(s): 0.6%us, 0.5%sy, 0.0%ni, 98.5%id, 0.1%wa, 0.0%hi, 0.3%si, 0.0%st
Mem: 3898468k total, 3558768k used, 339700k free, 111124k buffers
Swap: 4194300k total, 112k used, 4194188k free, 1352808k cached

Zombies:

USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
root 18256 0.0 0.0 0 0 ? Z 11:30 0:00 [aua.bin] <defunct>
root 18595 0.6 0.0 0 0 ? Z 11:32 0:00 [confd.plx] <defunct>



This thread was automatically locked due to age.
  • 0 in reply to Escape75

    Isn't that a pickle!

    If it's not the memory module, then the memory slot or memory controller.  Neither of which you can replace.

  • 0 in reply to Escape75

    Still, i'd be curious to the results of the prime95 test, the cpu torture and full ram. I bet one of those will crash the system.

  • 0 in reply to Jay Jay

    I will set it up to run from a bootable windows usb Slight smile

  • 0 in reply to Escape75

    Btw, I feel your pain in testing this.

    For the last 2 weeks been playing with *sense products.  I have att fiber which requires a special flavor of BS to work without their "required" gateway box.

    In a nutshell the service uses eapol on vlan0 to authentic and enable data flow.  This works great on utm (after installing wpa_supplicant), but *sense is freebsd based which has broken vlan0 implementations.  To get it working requires using something called netgraph to handle the vlan0 traffic at a kernel level.

    Anyway, there's multiple ways of implementing this, but some are claiming success without netgraph - this is welcome as fewer system resources are needed when transferring to internet at line speeds.

    Bottom line, I fully understand your frustration!

  • 0 in reply to Jay Jay

    Yes, that also sounds like a painful situation, but at least you're sharing the pain with all other pfsense users Slight smile

    What's wrong with using their "required" box and somehow turning it into a bridged unit?

  • 0 in reply to Escape75

    T's gateway box doesn't offer true bridge mode. Instead, it's an almost 1:1 NAT of sorts (certain ports are blocked) with the public ip getting passed to one device on the lan side. There's 4 lan ports on the rear. The gateway still offers connectivity to the other 3 ethernet on a 192.168.1.x/24 subnet.

    The theory is T is better able to mine user data with spyware in the gateway than doing it further upstream at the central office or NOC layer. Not to mention there's a much small state table (8K or 16K entries).

    Finally, the box is drawing 10-15 watts for doing absolutely nothing.

    Here's a good write up about T's auth process.  I am using the "supplicant" method.

    github.com/.../opnatt

  • 0 in reply to Jay Jay

    I see what you mean, it only supports IP Passthrough ...

    That reminds me of the time I had Shaw Cable and their modem didn't expose bridge mode,
    although it supported it technically, so one had to go into developer tools in a web browser
    and adjusting the code once logged in to the modem to re-enable the disabled functionality.

  • 0 in reply to Escape75

    Both the Small and Large FFT tests have passes, each running 2.5 hours ...

    The stick of RAM that was in the 'faulty' unit is also running fine in my r2 unit, I think I can pretty much exclude CPU, Memory and SSD ...

    Next I guess I'll test the interfaces ... 

  • 0 in reply to Escape75

    Does a known good stick work in the faulty unit? 

    Passes memtests and all?

Unfiltered HTML