BOINC Pentathlon 2019

[Gorgon]

1 hour ago, leadeater said:

Maybe I'm blacklisted or something. Earlier this morning I noticed it was having trouble giving out enough jobs to my systems, I do have a lot on a single account.

Nah I had a talk with your firewall admins and sent them a custom IPS regex just for you

[leadeater]

1 minute ago, Gorgon said:

Nah I had a talk with your firewall admins and sent them a custom IPS regex just for you

Firewall rule changes only happen after 4:30pm so I'm good till then 

[Gorgon]

 

Just now, leadeater said:

Firewall rule changes only happen after 4:30pm so I'm good till then 

That must lead for lots of overtime for the firewall team unless they have a night shift. If they do I hope they have at least one tier-2 person on it.

 

We just let them rip during the day except for a few things we do after hours.

 

They're talking about doing the scheduled change thing at work and if they do I'll probably take early retirement. Have to take 3 laptops home to work from home. Idiots in Security Architecture don't understand virtualization.

[leadeater]

1 minute ago, Gorgon said:

That must lead for lots of overtime for the firewall team unless they have a night shift. If they do I hope they have at least one tier-2 person on it.

It's just the push from Fortimanager, updates can happen any time just the push itself is after 4:30pm

[Gorgon]

I've killed Einstein and gone back to folding. I leave the Xeon running 7 threads on WCG for the Marathon and I have the two extra threads on each folding rigs running WCG but I'll have to keep an eye on that and make sure it doesn't impact folding too much but so far so good.

[Gorgon]

1 minute ago, leadeater said:

It's just the push from Fortimanager, updates can happen any time just the push itself is after 4:30pm

I know how that's supposed to happen but the reality is it can break shit left right & centre and it's better to have someone on-site to roll back that to try to figure it out the next morning with a line up at your door and insufficient caffeine. Different strokes ...

[Egg-Roll]

I'm letting my last Einstein process finish because I do like Pulsars  Beyond that all further tasks have been removed. All CPU's will be doing WCG.

 

Btw is there any way of getting a progress bar or even a simple % of the task at hand for Linux Command? Because seeing it would be nice. If not it's fine.

[Gorgon]

27 minutes ago, Egg-Roll said:

I'm letting my last Einstein process finish because I do like Pulsars  Beyond that all further tasks have been removed. All CPU's will be doing WCG.

 

Btw is there any way of getting a progress bar or even a simple % of the task at hand for Linux Command? Because seeing it would be nice. If not it's fine.

We should get this stuff worked out ahead of time next year.

 

Point boinc-manager at it from Windows. That way it might be quieter.

Following: https://boinc.berkeley.edu/wiki/Controlling_BOINC_remotely

edited /etc/boinc-client/remote_hosts.cfg adding <IP of Monitoring Station>
service boinc-client restart

[Gorgon]

Yeah, I'd say that F@H utilizes the GPUs better than Einstein

[leadeater]

1 hour ago, Gorgon said:

I know how that's supposed to happen but the reality is it can break shit left right & centre and it's better to have someone on-site to roll back that to try to figure it out the next morning with a line up at your door and insufficient caffeine. Different strokes ...

They are, the push is done at that time. I'm not even sure you can schedule it, without using API. We do it at that time because the firewalls are between server segments/subnets and it does reset sessions as it has in the past caused problems so during the day firewall changes is a hard no (even though it's not every time).

[Egg-Roll]

40 minutes ago, Gorgon said:

We should get this stuff worked out ahead of time next year.

 

Point boinc-manager at it from Windows. That way it might be quieter.

Following: https://boinc.berkeley.edu/wiki/Controlling_BOINC_remotely

edited /etc/boinc-client/remote_hosts.cfg adding <IP of Monitoring Station>
service boinc-client restart

Yea... I did a password my location was the one mentioned on the web page for the newer version. I'm not overly all worried, I find it annoying I can't see the progress threw a non-gui interface, esp since the one they give seems to have it in for me.

[Windows7ge]

52 minutes ago, Gorgon said:

We should get this stuff worked out ahead of time next year.

Way ahead of you. Plan to test a series of Linux distros both CLI & GUI then with virtualization I will be a bunkering god, lol.

[leadeater]

So these LHC WU seems to give stuff all credit, anyone know if the ones that use Virtual Box give much more?

[Windows7ge]

10 minutes ago, leadeater said:

So these LHC WU seems to give stuff all credit, anyone know if the ones that use Virtual Box give much more?

The work I've done hasn't shown any points yet but I can say from doing VM jobs in the past they take like 8hrs or more to complete and I think they only drop a few 1,000 or less per 8hr+ job.

[leadeater]

10 minutes ago, Windows7ge said:

The work I've done hasn't shown any points yet but I can say from doing VM jobs in the past they take like 8hrs or more to complete and I think they only drop a few 1,000 or less per 8hr+ job.

Might not be so bad on my big server, I'm concerned it's going to waste and would be better served on WCG since it gets huge output on that.

[Windows7ge]

2 minutes ago, leadeater said:

Might not be so bad on my big server, I'm concerned it's going to waste and would be better served on WCG since it gets huge output on that.

You have enough servers you could divy up the resources across multiple projects. We want to keep pressure on the marathon so if you think your bigger server would be better suited there I say go for it. Take a couple of the smaller ones and put them on LHC.

[leadeater]

2 minutes ago, Windows7ge said:

You have enough servers you could divy up the resources across multiple projects. We want to keep pressure on the marathon so if you think your bigger server would be better suited there I say go for it. Take a couple of the smaller ones and put them on LHC.

I have all but 1 on LHC, from what I know. Managing this many is actually not easy lol. I need to build out some scripts for next year, be much easier. Also should do an instance per project, minimum.

[Windows7ge]

2 minutes ago, leadeater said:

I have all but 1 on LHC, from what I know. Managing this many is actually not easy lol. I need to build out some scripts for next year, be much easier. Also should do an instance per project, minimum.

If that many machines are on LHC I'd put the big one back on Marathon. Gotta keep some pressure going there. I have ideas for next year but I'll need to go over them in practice first.

[leadeater]

3 minutes ago, Windows7ge said:

If that many machines are on LHC I'd put the big one back on Marathon. Gotta keep some pressure going there. I have ideas for next year but I'll need to go over them in practice first.

I WAS hoping to have an extremely good showing in LHC but..... this is not looking good. I have heaps of pending validation tasks and score just jumped up on the LHC website by 4k-5k so might leave it a few hours to see how it's going. If it's still looking like we're going to be destroyed in LHC I'll focus more on WCG again since I know we can pass AF on that.

[Windows7ge]

8 minutes ago, leadeater said:

I WAS hoping to have an extremely good showing in LHC but..... this is not looking good. I have heaps of pending validation tasks and score just jumped up on the LHC website by 4k-5k so might leave it a few hours to see how it's going. If it's still looking like we're going to be destroyed in LHC I'll focus more on WCG again since I know we can pass AF on that.

Going though the task list on the website I have a mass of pending validation tasks too. Explains why I have no credit yet. I'll find out probably tomorrow how much it's worth.

 

I would expect though if it's this hard for us it's just as hard for everyone else meaning it doesn't matter weather we continue to pressure LHC or go back to ZIKA. However I think LHC has jobs that take advantage of GPU acceleration so if teams get on that we might be hooped in which case going back to marathon would be our better bet.

 

I say check our placement for LHC tomorrow and if we're getting destroyed then maybe fall back on Marathon.

[leadeater]

3 minutes ago, Windows7ge said:

I would expect though if it's this hard for us it's just as hard for everyone else meaning it doesn't matter weather we continue to pressure LHC or go back to ZIKA

I was looking at other teams current total and hourly average and some are, like AF, 20 times ours.

[Windows7ge]

2 minutes ago, leadeater said:

I was looking at other teams current total and hourly average and some are, like AF, 20 times ours.

I'm pretty sure there are some LHC jobs that can use GPU so there's no direct way to combat that with our CPU's. If LHC is that much of a lost cause we'll have to fallback.

[Ithanul]

4 hours ago, Gorgon said:

Well this is a stupid competition IMHO. Just finishing Einstein on my GPUs and over 1/2 the WUs submitted are still "pending" and haven't been validated yet. Most of those have been submitted for at least 24 hours so their double validation is obviously broken in general or specifically due to the volume from the competition and people bunkering.

Yeah, no clue why folks voted for a quorum 2 project.  :I

 

 

3 hours ago, Gorgon said:

I know how that's supposed to happen but the reality is it can break shit left right & centre and it's better to have someone on-site to roll back that to try to figure it out the next morning with a line up at your door and insufficient caffeine. Different strokes ...

Yep, we do that where we work.  Any OS updates are after everyone is off the system and we stay till it is installed and tested.  Sometimes wind up with a 12 to 16 hour shift if something ain't playing right.

 

Going to be o so fun in the next months again.  Another OS update for us.

 

1 hour ago, leadeater said:

I have all but 1 on LHC, from what I know. Managing this many is actually not easy lol. I need to build out some scripts for next year, be much easier. Also should do an instance per project, minimum.

Yeah, I may play around with making some scripts as well when I do some testing after the event.

 

Edit:  Crap my brain doing word switching again.  Gah.

[leadeater]

17 minutes ago, Ithanul said:

Yeah, I make play around with making some scripts as well when I do some testing after the event.

Hoping to get rather fancy with them and put things in like target credits, auto bunkering after credits met, project/instance switching. Dunno, plenty of time to go crazy. Certainly would be very easy if you can just put in daily targets per project, project start and stop dates etc and it just takes care of itself.

[Ithanul]

Just now, leadeater said:

Hoping to get rather fancy with them and put things in like target credits, auto bunkering after credits met, project/instance switching. Dunno, plenty of time to go crazy. Certainly would be very easy if you can just put in daily targets per project, project start and stop dates etc and it just takes care of itself.

O darn, that sounds nice.  I be down to seeing that and/or helping.

 

I really need to get better at scripting.  Still a newbie at it.  Only been at my current job for one year so far.