On 05/02/2019 04.03, Dave Howorth wrote:
On Tue, 5 Feb 2019 02:31:17 +0100 (CET) "Carlos E. R." <> wrote:
El 2019-02-04 a las 15:37 -0000, Dave Howorth escribió:
On Mon, 4 Feb 2019 13:20:20 +0100 "Carlos E. R." <> wrote:
Whenever I try to debug problems like this I find it can be helpful to ssh into the machine with the problem as soon as it is started,
I tried. It did not connect.
Then investigate that problem first! If you boot a machine and cannot immediately ssh into it then there's something seriously wrong. (Unless it hasn't finished booting or sshd isn't enabled, of course). The whole point is to establish the ssh connection long before any problems arise.
You misunderstand.
No, you misunderstood. I made the original suggestion. You misunderstood it so I clarified and now you claim it's me that's misunderstanding! :( Cheeky!
You misunderstand again. In the paragraph you have removed the context: ]> You misunderstand. Of course I can connect normally when the computer is ]> normal. I can not connect when things are going on. ]> ]> You are saying to leave the ssh running for weeks till it happens? Uff. I ]> can not guarantee it. I show the evolution from misunderstanding to understanding, from the start of the paragraph to the end of it. It shows how initially I do not understand you, and later I do. I do not usually change the things I write as I write. Also, you misunderstood when I could not connect. All my machines can connect one to another. I said that "it did not connect" meaning that it hanged it could not connect.
Of course I can connect normally when the computer is normal. I can not connect when things are going on.
You are saying to leave the ssh running for weeks till it happens?
If it takes weeks for the problem to happen, yes.
Uff. I can not guarantee it.
It's difficult to understand why not. If the other machine needs to be rebooted for example, then it's perfectly possible to do that and reconnect the ssh session as soon as it's back up. The important point is to have the ssh session open at the time the primary machine demonstrates the problem.
We'll see.
Firefox is the default version that comes with it:
MozillaFirefox-60.4.0-lp150.3.30.1.x86_64
Desktop is XFCE.
FWIW, I'm using the same OS and FF but run LXDE.
I have two machines that started having problems after upgraded to 15.0.
Other possible variables are any add-ons you use in FF and the video hardware and drivers. I'd suggest running without add-ons, or at least only a few basic ones, and using the most basic video driver that works.
Happens on two very different computers. One laptop, one desktop. One Intel, other nvidia, one 4 gig, other 8 gig.
No, this is not hardware related. This is a fault of the kernel, how it mishandles swap into a dead embrace.
You've reported a kernel bug then, if you're so sure?
I'm sure, but I have no proof. Only that the process kswapd0 goes mad, about 25% CPU constant, on two machines. -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)