Untangle : Service Unavailable

[Justy]

I have observed something similar - and still opposite?
I has worked for months - 60+ Mbit throughput without any noticeable problems!

But this weekend, Untangle stopped routing traffic but I could still ping Untangle and access the web interface from LAN but nothing got routed through...

Eventually Untangle didn't give out DHCP responses so I guessed that Untangle was slowly falling apart. After a reboot it seems to work better but one machine still has problem getting through...

[sky-knight]

Swapping out NICs and randomly poking at an Untangle that throws this error will do you no good. I have a box here I picked up from a client that started throwing this error after I manually installed the thing to 7.0.1. No upgrade, just fresh install.

Guess what, the hard drive is bad. Just because the thing has run for months means exactly jack. Hardware dies, it just happens. Test the stuff and redeploy.

Once you've verified the hardware is good, and you're still getting this error. It essentially means, the UVM is offline. There are is no rack at this point, it may still be routing or bridging because that is all part of the linux kernel but the UT special stuff is gone.

The most common condition that causes this error is a corrupted UVM configuration database. The upgrade to 7.0 or 7.0.1 comes with a change in postgres version for most of us. That process isn't exactly clean. Factory reset your box and reconfigure by hand, do NOT restore a backup that will just put the problem DB back. Or, even better pave the box install a fresh 7.0.1 and reconfigure.

It is the only way to fix it.

[pod]

The most common condition that causes this error is a corrupted UVM configuration database. The upgrade to 7.0 or 7.0.1 comes with a change in postgres version for most of us. That process isn't exactly clean. Factory reset your box and reconfigure by hand, do NOT restore a backup that will just put the problem DB back. Or, even better pave the box install a fresh 7.0.1 and reconfigure.

It is the only way to fix it.

I'd like to hear this officially from someone at Untangle, if it's true.

I'd also like to know where to look in the logs to figure out exactly what is going on with UT and, ideally, how to fix it.

I understand what you're saying about disks possibly being broken, but this one at least isn't logging any disks errors, so I find that hard to believe. I'm willing to be persuaded though.

[sky-knight]

This one here isn't either... factory diagnostics on this Seagate SATA drive are perfect. Yet, no matter what I do to it, on restart it loses data. I think the drive cache has failed. Normally I would blame the mainboard's controller, but replacing the drive appears to have resolved the issue for this one.

[Gremlin]

I'm sitting on a site in bridge mode with 1700 computers behind it and we are getting the same "Service Unavailable" message when trying to log in the web interface from the local box. From my office computer we get nothing, it just times out. We got it back up by starting the backup/restore utility and backing out of it (restarts the UVM). We only noticed this after the 7.0 to 7.01 upgrade. Our drives are fine and everything works once the UVM is restarted.

I'm not sure we have the exact same issue but it appears we all are experiencing some failure of the UVM after the 7.01 upgrade. From a tech view of what changed, I assume something the .01 patch did wasn't kosher. Does anyone know exactly what changed?

Sky-Knight, it is my understanding that the Seagate diagnostic utility checks the drive platters and the on-board circuitry (cache, drivers, etc). As far as a fresh install then reconfiguring... That doesn't seem like much of a solution for someone in my shoes.


Edit: Ok I own sky-knight an apology. Our Seagate HD is dead, really really dead. Did the Seagate test suite and it didn't pass a single test >_<, sorry.

[sky-knight]

I don't have an easy answer for you, it wasn't the 7.0.1 upgrade that was the problem, it was the 7.0 upgrade in general. You simply cannot swap out the DB server on a box and expect it to work all the time. We had similar issues when we made the leap from 5.2 to 6.2.

Never have automatic upgrade configured, always have a backup configuration and the previous version's installation media handy in case something goes wrong. You can't just fix a DB, it must be restored, and because you don't have a working DB of that version, the only option is reset to defaults / reconfigure, or pave / reconfigure.

[Gremlin]

We are a paying site so we have the auto backup plus I like keeping a daily copy locally, but from what I read that isn't going to help us as we need to go back to 6.2? Then we can't update?

Or are you saying wiping the database (factory reset) then restoring from the backup will put everything where it needs to go? Also if the backup is from 7.01 and it has issues won't restoring the back just reintroduce the same problems (not sure how Untangle backups)? I am also aware that 7.0 had some issues with restoring from backup; did that include creating backups?

[sky-knight]

Correct, you need a working backup from 6.2, and restore to a 6.2 installation.

Then you can attempt the upgrade again... it doesn't always go wrong, it just isn't 100% right either.

Then again if you're a paying site you could just put in a call to support. They might be able to fix it.

[otis]

Sky-Knight you raise a valid potential cause and in some cases that could be it however...

Is it likely that all of us reporting this issue would have HDD issues around the same time? Possible yes, likely no.

This seems more likely a code issue causing the service failure.
In my case if the service fails data still passes through and no filtering takes place, when the service is restarted the entire process resumes function normally. The only data loss seen is during the time the service not running.

This is not an upgrade. This is clean install of 7.0.1 with newly created DB.
I should add that potentially this started on 7 (not 100% my original 7 had not upgraded to 7.0.1 yet) but I thought it may have been related to something else I did and went to 7.0.1 to get the same results.

[sky-knight]

Otis in that case you have something going wrong that is causing the UVM to actually crash. In my experience the only two things to consistently cause this are defective hard drive or defective ram.

Then there are the software issues from the upgrade, which can cause a corrupt database.

I believe the log file with relevant information on the crash would be /var/log/uvm/uvm.log

But I'm far from a master at decrypting that stuff. I usually refer this type of thing to Untangle support.

I can assure you the code is working in this regard. There are thousands of Untangle installations out there, this thread is working with what? 5? I have 7.0.1 installations in the field now working just fine.

So the question is what is different with your installs.