Werner changed the topic of #armbian to: armbian - Linux for ARM development boards | www.armbian.com | Github: github.com/armbian | Commits: #armbian-commits | Forums Feed: #armbian-rss | This channel is logged -> irc.armbian.com
Toast has quit [Ping timeout: 246 seconds]
sunshavi has quit [Ping timeout: 250 seconds]
Tonymac32 has joined #armbian
<Tonymac32>
so, still trying to pull meaningful debug info off of my Odroid MC1's, but
<Tonymac32>
remember that overflow bug that made booting fail?
<Tonymac32>
2 of my MC1's were immune to it it seemed
<Tonymac32>
and now, running 5.4.19 and Einstein@home, the same 2 of them run fairly consistently, the other three are crashing on disk access
<Tonymac32>
knocking down the clock speed helped (I am keeping them cool and have a 300W supply I have been monitoring voltage)
<Tonymac32>
lanefu did you get some of these?
<lanefu>
Tonymac32: yeah got 4 of 'em
<lanefu>
hmm lemme check on my uptime
<lanefu>
surprise i crashed prometheus or somethign trying to query big lol
<lanefu>
looks like longest uptime i've had is 2 weeks
<lanefu>
tends to be a lot of factors as to why i reboot
<Tonymac32>
with light loads I had 28 days uptime
<Tonymac32>
the heavy 100% loads ran stable for duration, but upon completion the machines faceplanted
<lanefu>
ohh so it freaked out going idle?
<Tonymac32>
I'm not certain. the logs aren't giving many secrets
<Tonymac32>
to give a hint as to the insane variation in stability, each workload from the server is worth 63 points. one of the boards has only managed to upload 1 completed workload, while another has completed 12
<Tonymac32>
same supply running all
<lanefu>
let me find that ssl command taht somebody had me use to stress test an h6
<lanefu>
it was amazing
<lanefu>
would just lock it up so fast
<sarnold>
openssl speed?
<sarnold>
perhaps with -multi to knock up concurrency? openssl speed -multi +8
<Tonymac32>
just like the atrocity known as the Rockchip kernel supports the cameras and vdec
<Tonymac32>
I read Amazon will only be delivering important things
<lanefu>
yeah so they're starting with controling what goes into their warehouses
<lanefu>
and then prioritize out
<Tonymac32>
right
<lanefu>
its a good thing I have drawers full of good intentions
<lanefu>
shelves really
<Tonymac32>
I grew up poor in the country, I have always kept a few week's worth of food handy, and bought in bulk to save
<Tonymac32>
I might also have 30 or so pints of green beans, etc, canned fromt he garden.
<Tonymac32>
...and a gallon of rumtopf I forgot about around Christmas since it was such an ordeal
<Tonymac32>
:-D
<lanefu>
haha that'll keep you hydrated
<Tonymac32>
quite conveniently I also have some vanilla cream soda
<lanefu>
i've got a lot of small markets nearby and they dont really get beatup like the bigbox places
<lanefu>
and i don't have kids
<lanefu>
so my wife adn I have it really easy in this situation
<lanefu>
are you WFH now? or you still have to go in like a hardworking american
<Tonymac32>
lol I work for a German company, we are almost mandatory work from home now
<Tonymac32>
until next month so far
<lanefu>
yeah i work for a french company they gave the order on friday
<lanefu>
wife works for financial company that has a huge campus here. they're all remote now
<lanefu>
they had the brains to do a large scale VPN capacity test last monday
<lanefu>
schools are closed
<Tonymac32>
same
<lanefu>
been doing strange things like I cleaned my office
<Tonymac32>
haha
<lanefu>
so yeah influxdb swallowed all my ram
<Tonymac32>
lol
ced117 has quit [Ping timeout: 256 seconds]
ced117 has joined #armbian
ced117 has quit [Changing host]
ced117 has joined #armbian
sunshavi has joined #armbian
<Tonymac32>
I'm going to try to run the data files on USB
xec has joined #armbian
xecuter has quit [Ping timeout: 246 seconds]
dddddd has quit [Ping timeout: 246 seconds]
maccraft has joined #armbian
maccraft has quit [Ping timeout: 264 seconds]
<IgorPec>
lanefu:
<ArmbianTwitter>
@raspberrytorpes (raspberryparatorpes): RT @Khomutoff: #PiDay project: All add-free internet at home with #PiHole! Just add #orangepi, #armbian and some very light system engineer⦠(13s ago)
archetech has quit [Quit: Konversation terminated!]
archetech has joined #armbian
maccraft has quit [Ping timeout: 246 seconds]
maccraft has joined #armbian
<Tonymac32>
IgorPec have you seen any instability with XU4 under load?
<Tonymac32>
5.4 kernel
<IgorPec>
5.4?
<IgorPec>
huh, not ... but i have seen more patches upsream addressing this
<Tonymac32>
1 out 5 machines will run more than a few hours with a BOINC instance
<Tonymac32>
0
<IgorPec>
i got one machine in my great setup, but currently still working on the system
<Tonymac32>
haha
<Tonymac32>
So oddly enough 2 of 5 are way more stable than the others
<Tonymac32>
and they would boot the kernel even without that overflow patch
<Tonymac32>
1 of them is on 24 hours up time as of now, the rest have been unplugged until I can look more closely
<Tonymac32>
part-part variation like that makes me think RAM or SD interface
<IgorPec>
yeah, i apparently there is something still wrong
<IgorPec>
ram or perhaps also dvfs?
<Tonymac32>
possibly, but I knocked them all down to 1.6 GHz which seemed to help, but to your point some of the new upstream stuff is a lot of clock work
<IgorPec>
try setting some fixed speed and crunch numbers ;)
<Tonymac32>
yeah, I'll try that
<Tonymac32>
we do have some dmesg anger concerning a couple of the opps
<Tonymac32>
maybe it trips on one
<IgorPec>
yeap, do some searching on kernel mainline lissts. i think i saw a recent conversation on this topic there
<IgorPec>
but can't find it now
<Tonymac32>
I thought BOINC would be a good stress test, I was correct. all 5 machines were up for 28 days with lighter loads before this
<Tonymac32>
:P
<IgorPec>
aha, so heavy stressing kills them quickly, right?
<Tonymac32>
well, they seem ok while the load is on, but somehthing is happening when a work packet is complete, either a filesystem access, or a lot of RAM activity to "close out" and finalize, then it dies
<Tonymac32>
so it takes roughly 5 hours to put in a work package, all is ok for that time, but you hit 99.95% complete, and then *BOOM*
<Tonymac32>
the thing just shuts off
<Tonymac32>
heartbeat LED stuck in whatever state it was in when the thing died
<ArmbianTwitter>
@LlnuxBot (LinuxBot): RT @XGRldlxpY2UK: New on premise cluster #Pine64 clusterboard with #Armbian #Linux Quad-Core ARM Cortex A53 and 2GiB LPDDR3 RAM per slot,β¦ (12s ago)
drobo_00 has joined #armbian
<ArmbianTwitter>
@armbian (armbian): RT @XGRldlxpY2UK: New on premise cluster #Pine64 clusterboard with #Armbian #Linux Quad-Core ARM Cortex A53 and 2GiB LPDDR3 RAM per slot,β¦ (12s ago)
<ArmbianTwitter>
@Poddingue (Bruno Verachten π°): RT @XGRldlxpY2UK: New on premise cluster #Pine64 clusterboard with #Armbian #Linux Quad-Core ARM Cortex A53 and 2GiB LPDDR3 RAM per slot,β¦ (28s ago)
<ArmbianTwitter>
@LinuxDreams (LinuxDreams): RT @XGRldlxpY2UK: New on premise cluster #Pine64 clusterboard with #Armbian #Linux Quad-Core ARM Cortex A53 and 2GiB LPDDR3 RAM per slot,β¦ (8s ago)
Toast has quit [Quit: Toast]
<ArmbianTwitter>
@worksonarm (Works on ARM): RT @XGRldlxpY2UK: New on premise cluster #Pine64 clusterboard with #Armbian #Linux Quad-Core ARM Cortex A53 and 2GiB LPDDR3 RAM per slot,β¦ (10s ago)
Toast has joined #armbian
<ArmbianTwitter>
@sskras (Saulius Krasuckas): RT @XGRldlxpY2UK: New on premise cluster #Pine64 clusterboard with #Armbian #Linux Quad-Core ARM Cortex A53 and 2GiB LPDDR3 RAM per slot,β¦ (7s ago)
maccraft has quit [Ping timeout: 256 seconds]
maccraft has joined #armbian
archetech has quit [Quit: Konversation terminated!]
<Tonymac32>
ok, brought one up at 1.8 GHz fixed, let's see what happens
<IgorPec>
watch out that it doesn't burn :)
maccraft has quit [Ping timeout: 250 seconds]
maccraft has joined #armbian
xec has quit [Remote host closed the connection]
<Tonymac32>
a cool 60C, those aluminum brackets appear to be well bonded
xecuter has joined #armbian
maccraft has quit [Ping timeout: 256 seconds]
xecuter has quit [Remote host closed the connection]