Hey Everyone,

The server appears to be having some connectivity issues. Additionally, one of the drives on the server appears to have failed - luckily things are mirrored so we haven't lost any data. However, I need to make a backup of things, which could take several hours. Once everything is backed up I'll try to clear out this transitioner/validator backlog and get things up to speed again.

Best,
Tom

The problem with the bad drive has been handled, but things are struggling to come back up. I'm working on it currently.

Thanks a lot Tom. Really appreciate you letting us know when things are not running properly.

Thanks a lot Tom. Really appreciate you letting us know when things are not running properly.

Double for me. I've a few that need to come back home..

Things are looking up. I was able to send in 47 tasks and get one back for my slowest cpu.

Looks like things are stable for the moment, I'm going to keep an eye on the validator and transitioner backlogs. Hopefully they will start decreasing as things begin to flow again.

I also think the server is running slower than usual because it is re-configuring itself after losing the broken hard drive.

We will have to take the server down again in a few days to replace the bad drive, but I'll let you know when that happens. I'll also keep in touch in case any more maintenance is needed before then.

Tom,

No worries. Have you tried hitting it with a very large hammer? :)

Seriously though, all of us "data crunchers" truly appreciate all of the infrastructure work and support that you provide.

Kind Regards,
Cliff
Philadelphia, PA

Thanks for the hard work Tom.
One of the benefits of the 14 day deadlines is the results can have a few days till they're returned if something unexpected happens.

Hey Everyone,

The server appears to be having some connectivity issues. Additionally, one of the drives on the server appears to have failed - luckily things are mirrored so we haven't lost any data. However, I need to make a backup of things, which could take several hours. Once everything is backed up I'll try to clear out this transitioner/validator backlog and get things up to speed again.

Best,
Tom

When I ran servers we had RAID 6. Yes mirroring isn't exactly the same, but surely you can just pull out the bad drive while it's running and shove in another and it rebuilds it in the background? I even used this function to upgrade to larger disks without interrupting the users, I just did one at a time.

Another question, sorry. How come whenever it recovers from a problem, the server ends up with 2 million tasks queued to send out instead of the usual 11000?

Anyway it was nice to walk into my garage this morning to feel the temperature of a butterfly house. I knew what that meant! My Amazon parrots thank you.

Time to break out the old Tonia Harding and giver her a good whack.......

Time to break out the old Tonia Harding and giver her a good whack.......

Is that Cockney Rhyming Slang?
I had to look up who she is, my god how can a cute thing become so repulsive?

Luciferius Infernalis Vel Tohu

My Latin isn't so good. Infernal devil?

Thanks

---

Another question, sorry. How come whenever it recovers from a problem, the server ends up with 2 million tasks queued to send out instead of the usual 11000?

This can be due to connection issues, so the server keeps accumulating jobs but isn't able to establish connections to volunteers to hand them out. It can also be due to server memory problems, where it doesn't have the memory available to query the database fast enough to hand everything out in time.

Another question, sorry. How come whenever it recovers from a problem, the server ends up with 2 million tasks queued to send out instead of the usual 11000?

This can be due to connection issues, so the server keeps accumulating jobs but isn't able to establish connections to volunteers to hand them out. It can also be due to server memory problems, where it doesn't have the memory available to query the database fast enough to hand everything out in time.

I thought you'd set a limit, as it's usually precisely 10000 seperation and 1000 Nbody.

I believe the server tries to keep at least that many tasks queued up to be ready to send out, but sometimes it increases if the server is struggling.

Thanks for all you do

Server not fixed yet?

There is a large number of workunits that are waiting to be validated. I think that this is because the server is running more slowly while it is degraded. When we get the drive back we can restore full functionality of the server. Hopefully that isn't too long, but in the meantime this big drop in RAC might just be how things are at the moment.