BOINC Pentathlon 2020

[Gorgon]

Just now, Windows7ge said:

I already had the whole thing apart while dusting it out. The RAM isn't socketed.

 

If it ever boots again I'll consider a tftp server. For now I'll try the firmware update and see if that gets me anywhere.

If the flash is socketed you can often read it on a PC as most is formatted with FAT. Good luck finding a 256MB or so Compact Flash though if it is a corrupt Flash

[Windows7ge]

3 minutes ago, Gorgon said:

If the flash is socketed you can often read it on a PC as most is formatted with FAT. Good luck finding a 256MB or so Compact Flash though if it is a corrupt Flash

Everything is soldered. Though I do know my way around a hot-air re-workstation fairly well. I wouldn't know where to source a new chip or the software for it though.

[Gorgon]

3 minutes ago, Windows7ge said:

Everything is soldered. Though I do know my way around a hot-air re-workstation fairly well. I wouldn't know where to source a new chip or the software for it though.

Yeah, likely easier just to buy another one from eBay.

 

I'm using Cisco Small Business 200-series switches at home. They're re-branded LinkSys and I miss the CLI but they're reliable and have the features I wanted (VLANs, Link Aggregation, SNMP) for a reasonable price but Lord, do I hate the GUI on them.

 

We use exclusively Cisco at work now. We used to use Nortel for layer2 but they're slowly being replaced as they die out. We still have some Bay Networks 450-24T switches in production I installed back in 1996 when they were new.

[leadeater]

22 minutes ago, Gorgon said:

I'm using Cisco Small Business 200-series switches at home. They're re-branded LinkSys and I miss the CLI but they're reliable and have the features I wanted (VLANs, Link Aggregation, SNMP) for a reasonable price but Lord, do I hate the GUI on them.

I'm using an SG300, wanted the static routing, but yea the GUI isn't the best but also not the worst I've seen either lol.

[Gorgon]

21 minutes ago, leadeater said:

I'm using an SG300, wanted the static routing, but yea the GUI isn't the best but also not the worst I've seen either lol.

I keep wanting to go 10GigE at home but don't have the use case. Maybe when 2.5G is more developed. I have a 24-port switch and 3 8-ports and I'm running out.

[leadeater]

FYI

 

Tasks in progress at announcement time was less than 100k, think there are going to be some big first throws maybe.

 

Edit:

Now 345k

[Windows7ge]

@Gorgon The flash appears to have been succesful. "show bootvar" now reports 2.0.0.68. We did get further this time as well, more reliably too however it still seems to want to crash or fail to boot. I got it to run without much hassle but it spit out this error:

01-Jan-2000 00:00:17 %2SWVLAN-F-VLNADDMAC: SW2P_vlan_add_vlan_tome_mac_addr: Fai
led to add mac_to_me entry, vlan ? - error code SW2P_vlan_add_vlan_tome_mac_addr


***** FATAL ERROR *****
Reporting Task: BRMN.
Software Version: 2.0.0.68 (date  02-Dec-2018 time  16:53:06)
0x188d9c
0x18637c
0x5b8b44
0x3e59d8
0x3e9450
0x7870d4
0x789860
0x65d24c
0x65db88
0x65dc2c
0x532214
^^ this event was logged before the last reboot.

***** END OF FATAL ERROR *****

I'm going to leave the switch to run for a little while. See if it reboots on me.

 

Edit:

Another error.

01-Jan-2000 00:00:11 %OS-F-ITC: OSITCG_sem_p: Kernel failed: sem_id = 0, RTOS re
turned status 6

 

[Gorgon]

29 minutes ago, Windows7ge said:

@Gorgon The flash appears to have been succesful. "show bootvar" now reports 2.0.0.68. We did get further this time as well, more reliably too however it still seems to want to crash or fail to boot. I got it to run without much hassle but it spit out this error:

01-Jan-2000 00:00:17 %2SWVLAN-F-VLNADDMAC: SW2P_vlan_add_vlan_tome_mac_addr: Fai
led to add mac_to_me entry, vlan ? - error code SW2P_vlan_add_vlan_tome_mac_addr


***** FATAL ERROR *****
Reporting Task: BRMN.
Software Version: 2.0.0.68 (date  02-Dec-2018 time  16:53:06)
0x188d9c
0x18637c
0x5b8b44
0x3e59d8
0x3e9450
0x7870d4
0x789860
0x65d24c
0x65db88
0x65dc2c
0x532214
^^ this event was logged before the last reboot.

***** END OF FATAL ERROR *****

I'm going to leave the switch to run for a little while. See if it reboots on me.

 

Edit:

Another error.

01-Jan-2000 00:00:11 %OS-F-ITC: OSITCG_sem_p: Kernel failed: sem_id = 0, RTOS re
turned status 6

 

Did you wipe the config?

 

Once you get in to the CLI first type "enable" to elevate your privileges then

console# delete startup-config

to reset to factory defaults. You likely have to:

copy run start

to save the default config

[Windows7ge]

23 minutes ago, Gorgon said:

Did you wipe the config?

 

Once you get in to the CLI first type "enable" to elevate your privileges then

console# delete startup-config

to reset to factory defaults. You likely have to:

copy run start

to save the default config

If it already has a running config won't running "copy run start" just replace the startup-config with the configuration that's already in mem? Think I'd have to do a restart prior to copy run start no?

[Gorgon]

14 minutes ago, Windows7ge said:

If it already has a running config won't running "copy run start" just replace the startup-config with the configuration that's already in mem? Think I'd have to do a restart prior to copy run start no?

It depends on if the factory default command wipes the startup on flash or jus the one in memory.

 

Try without the save and if it doesn’t work try it with the copy running startup

[Windows7ge]

29 minutes ago, Gorgon said:

It depends on if the factory default command wipes the startup on flash or jus the one in memory.

 

Try without the save and if it doesn’t work try it with the copy running startup

Quote

Welcome to Dell Easy Setup Wizard.

I think it worked.

 

At this time I am not seeing any more error messages.

 

The switch has stopped resetting.

 

No more kernel panics.

 

I plugged a couple devices into it and it's working. Holy jeez, we may have just fixed it for real, lol.

 

So it was a firmware update & factory reset that did it. It's 11:00PM now gonna hang it up for the night but in the coming days I'll swap out the home switch for this one and see if it can handle a modest load without complaining. If it does it's going in my server rack.

[Gorgon]

14 minutes ago, Windows7ge said:

I think it worked.

 

At this time I am not seeing any more error messages.

 

The switch has stopped resetting.

 

No more kernel panics.

 

I plugged a couple devices into it and it's working. Holy jeez, we may have just fixed it for real, lol.

 

So it was a firmware update & factory reset that did it. It's 11:00PM now gonna hang it up for the night but in the coming days I'll swap out the home switch for this one and see if it can handle a modest load without complaining. If it does it's going in my server rack.

Sometimes it's like BIOS settings after loading a new BIOS need to be cleared. Could have been an old vlan config for the VoIP phones causing confusion or some command not supported in the new firmware version.

[leadeater]

3 minutes ago, Gorgon said:

Sometimes it's like BIOS settings after loading a new BIOS need to be cleared. Could have been an old vlan config for the VoIP phones causing confusion or some command not supported in the new firmware version.

If it is on the same OEM/ODM hardware that HP used then corrupt firmware and flash is very common, we got it literally all the time on 5820's and was never really fixed until 5830's. Think it was that model series anyway, was 5x20 what ever that x actually is.

 

Re-flash and wipe required every time.

[Windows7ge]

34 minutes ago, Gorgon said:

Sometimes it's like BIOS settings after loading a new BIOS need to be cleared. Could have been an old vlan config for the VoIP phones causing confusion or some command not supported in the new firmware version.

There was this one error message explicitly mentioning VLANs:

01-Jan-2000 00:00:17 %2SWVLAN-F-VLNADDMAC: SW2P_vlan_add_vlan_tome_mac_addr: Fai
led to add mac_to_me entry, vlan ? - error code SW2P_vlan_add_vlan_tome_mac_addr

But I think the problem comes before that since it was previously crashing while decompressing the image. The OS hadn't even booted yet let-alone start-up the VLANs. I think this issue with the VLAN was a consequence of something deeper that went wrong.

[RollinLower]

quick question: i'll be doing the marathon this year on all my systems. would rosetta benefit from using the GPU's aswell?

[leadeater]

2 minutes ago, RollinLower said:

quick question: i'll be doing the marathon this year on all my systems. would rosetta benefit from using the GPU's aswell?

Nope it's CPU only, so you can do other GPU events or FAH concurrently.

[Gorgon]

1 hour ago, RollinLower said:

quick question: i'll be doing the marathon this year on all my systems. would rosetta benefit from using the GPU's aswell?

The Javelin event supports GPUs

[marknd59]

What's the plan for the Javelin tomorrow ? I've got 1 system that has been bunkering Rosetta and Numbers GPU tasks for the last couple of days that I will upload tomorrow and I'll leave it doing that all day. My other system has been bunkering Rosetta which I'll also upload tomorrow, but it has 2 x RTX 2060's in it that are currently doing F@H Wu units.

 

Are we making a push for the Javelin tomorrow and should switch my 2060's over to Numbers after I upload all the Rosetta tasks or are we letting the Javelin slide tomorrow and making a push for it on the other days ?

[ffbr]

My system crunchin task for numbers ran out and I do not seems to be able to pick new ones. Any ideas?

[leadeater]

11 minutes ago, marknd59 said:

Are we making a push for the Javelin tomorrow and should switch my 2060's over to Numbers after I upload all the Rosetta tasks or are we letting the Javelin slide tomorrow and making a push for it on the other days ?

I would wait out the first throw and see how things land, right now we have no idea what we need or what the other teams are doing and personally I have no idea what a good RAC/PPD is for Numbers so I'm flying blind. Just go with what we have for the first one then we'll get an idea of how things look, not that it would hurt to switch those GPUs to Numbers if you wanted to.

 

I've got a really large stockpile of Number jobs myself, majority were done on CPU, but I actually don't know how much I have. Think I found the limits of boincview, it can't handle the number of systems and tasks I have and it just locks up, I know I should have at least 10,000 completed Numbers jobs so maybe like 1.2 mil points?????

[leadeater]

3 minutes ago, ffbr said:

My system crunchin task for numbers ran out and I do not seems to be able to pick new ones. Any ideas?

What does the log say? Maybe just run up a new instance of boinc on the same system.

 

https://boinc.berkeley.edu/forum_thread.php?id=10986

https://www.overclock.net/forum/18056-boinc-guides-tutorials/1628924-guide-setting-up-multiple-boinc-instances.html

[ffbr]

2 minutes ago, leadeater said:

What does the log say? Maybe just run up a new instance of boinc on the same system.

 

https://boinc.berkeley.edu/forum_thread.php?id=10986

https://www.overclock.net/forum/18056-boinc-guides-tutorials/1628924-guide-setting-up-multiple-boinc-instances.html

Well ... didn't think to look at the log...

So yes, numbers tell me to back off  "too many upload in progress"

Gonna set up another boinc instance

[ffbr]

This is highly frustrating, my T4s also ran out of job during the night, such a waste. Glad the V100 are currently not available, they would probably ran out of jobs in 1h

 

 

[ffbr]

12 hours ago, miker07 said:

Just an fyi I've go about 70 cores (140 threads) on marathon and  a few gpus (2 radeon vii's, 3 2080ti's, and a 2060 super) on javelin, and started bunkering today. Sounds like universe for city run runs significantly better on linux, so since mine are all windows I don't plan to run unless requested. With numberfields, it isn't pushing the gpus at all, does anyone know if there are config settings to fully utilize gpu's, or can it not push them? I know sometimes you can try running multiple tasks per gpu on some projects, but that isn't recommended on some gpu projects. Collatz likes 1 per gpu, while einstein and milkyway like multiple tasks per gpu. 

Wonder if BOINC run on the WSL. Anyone have tried it?

 

[ffbr]

What does the quorum means for the projects?