Random crashes after upgrade to 16.6.2 from 16.5.2

[oj88]

I hope v17 fixes more problems than it introduces.

My clean-installed 16.6.2 out of the blue rebooted once about 9 days ago. I haven't changed anything and have kept IPS enabled. So far it hadn't reoccurred.

[bcarmichael]

We've identified the performance issue with IPS and we will be fixing it in the next release of version 17 which we will publish soon. Thanks for your feedback.

[darthrater]

I had this same issue with 16.6..2 and the tech "support" was laughingly, insultingly bad. Also blaming my issue on "hardware". It's a VM, you muppet. So sky-knight thanks so much for your insight here. I just upgraded again with IDS off and my resources also are not haywire like before.




In terms of support griping , just last week I called for a reports issue, the guy ran a command on CLI and ended up completely breaking my entire install, so I had to restore from a backup. He couldn't have cared less.

[petecloss]

I've spent the last day and a half tearing my hair out (including hours spent on associated trouble with my Ubiquiti switches and APs) before finally plugging a monitor into my Untangle router and noticing it rebooting periodically. I've now disabled IDS and it's been stable ever since. Reinstalling and re-enabling IDS caused a hard reset within seconds, so that hasn't proven a solution to this issue for me.

I likewise upgraded to 16.6.2 from 16.5.2 yesterday, I'm running Home Protect Basic on an HP T730 with a 250GB Samsung 860 EVO M.2 SSD (bought October 2021). Multiple Memtest86+ runs passed without errors.

[sky-knight]

You can see in the screen grab here, my upgrade from 16.5 to 16.6, the reboot, then the CPU going bonkers after things got spun up... then me turning off IDS and it calming down, then me removing IDS, reinstalling it, and re-configuring it. IDS is online now, and CPU is back to normal.

So if your CPU went nuts post upgrade, I suggest reinitializing the IDS module, it seems to have cured my unit. It's still using more CPU, but it's in the category of what I'd expect for upgrades over time, not orders of magnitude like it started.

[agit8or]

I was just asked by tech support to share this thread. I guess they dont even monitor their own forums anymore.

[agit8or]

After spending a week with tech support (Which is now almost worthless) blaming it on hardware, we tried disabling IDS on 4 higher traffic boxes we were constantly having issues with. That was 5 days ago, zero crashes. Do better for your partners Arista.

[agit8or]

I made a post about this and it was censored. We have multiple boxes randomly rebooting and UT blames it on the hardware, yet they have zero issues before the upgrade. Tech support has gone downhill.

[road hazard]

Your situation is completely different, your hardware is fine just the drivers are suspect because Dell meets Debian, that's just the way it goes. You can fix your hardware AND be supported by using VMWare. The OP here has Protectli junk and most likely his SSD gave out during upgrade, because that's what they do.

Hang around Debian long enough, doesn't matter who's product uses it, and you'll see these same patterns.

Good point (regarding the different hardware in use).

I put ESXi on my server a few days ago and testing things out with spinning up Linux and Windows workstations and so far so good. Been doing some disk intensive bench-marking in the Windows VM and after a bit over 48 hours, no spontaneous reboots. Still keeping a watchful eye out for v17 .

[sky-knight]

Experienced the same thing. Fresh install of 16.6.2 and had a random crash. Made a post about it over here: https://forums.edge.arista.com/forum...taenous-reboot

When v17 drops, I'm going to run it in ESXi and see how it goes.

Your situation is completely different, your hardware is fine just the drivers are suspect because Dell meets Debian, that's just the way it goes. You can fix your hardware AND be supported by using VMWare. The OP here has Protectli junk and most likely his SSD gave out during upgrade, because that's what they do.

Hang around Debian long enough, doesn't matter who's product uses it, and you'll see these same patterns.

[road hazard]

Experienced the same thing. Fresh install of 16.6.2 and had a random crash. Made a post about it over here: https://forums.edge.arista.com/forum...taenous-reboot

When v17 drops, I'm going to run it in ESXi and see how it goes.

[sky-knight]

Well, when looking at these system stats, they don't show any strain at crash/reboot time. FWIW Hardware is Protectli Vault FW4B - 4 Port, Intel Quad Core, 8GB RAM, 120GB mSATA SSD

That platform? that mSATA is dying... they love doing that. One of the many reasons I refuse to sell China junk like Protecli does. Here, you can import them yourself: https://www.alibaba.com/trade/search...rity+appliance

If you're going to buy garbage, you may as well get it cheap. The recent upgrade just pushed that garbage drive over it's wear endurance level, because it's basically that of a Micro SD card.

[DCROK]

Intrusion Prevention is very memory and disk intense. It indicates the hardware is either buckling under the strain, or the new kernel has an "updated" driver that isn't handling the strain. Specifically, it would be a chipset or storage problem. Could be memory too, but that's a stretch.

Well, when looking at these system stats, they don't show any strain at crash/reboot time. FWIW Hardware is Protectli Vault FW4B - 4 Port, Intel Quad Core, 8GB RAM, 120GB mSATA SSD

[sky-knight]

Intrusion Prevention is very memory and disk intense. It indicates the hardware is either buckling under the strain, or the new kernel has an "updated" driver that isn't handling the strain. Specifically, it would be a chipset or storage problem. Could be memory too, but that's a stretch.

[DCROK]

fyi - I turned off Intrusion Prevention app and system has been running without random reboots for 5 days now. No real solution but at least we know what's the problem.