Shouldn't the non-ghosts go into "urgent" mode (can't think of the right term right now) when the deadline approaches ? I don't see the necessity to run the cache dry.

I don't think there is a high probability to miss the deadline if the computer is crunching on a regular basis because the deadline is pretty long compared to the possible cache size. Of course if someone would download a 10 day cache, and then only turns on the client for an hour a day, that might cause a problem.

As far as the ghosts go, I just did the math on my account info, and it looks like I am holding steady at about 1500 ghosts at the moment, which means the new ghosts I am getting are counterbalanced by the ghosts timing out. I really have no problem with that, and eventually SETI will solve the ghost problem without me having to mess around with it.

The only suggestion to prevent ghosts that I saw so far came from Arne, I believe. He suggested to wait with the network communication until Sunday or even Monday, and by that avoiding the problem which creates ghosts in the first place. It sounds logical to me, but I personally don't want to take the risk of not getting enough work to last through the server downtime.

---

Shouldn't the non-ghosts go into "urgent" mode (can't think of the right term right now) when the deadline approaches ? I don't see the necessity to run the cache dry.

I don't think there is a high probability to miss the deadline if the computer is crunching on a regular basis because the deadline is pretty long compared to the possible cache size. Of course if someone would download a 10 day cache, and then only turns on the client for an hour a day, that might cause a problem.

As far as the ghosts go, I just did the math on my account info, and it looks like I am holding steady at about 1500 ghosts at the moment, which means the new ghosts I am getting are counterbalanced by the ghosts timing out. I really have no problem with that, and eventually SETI will solve the ghost problem without me having to mess around with it.

The only suggestion to prevent ghosts that I saw so far came from Arne, I believe. He suggested to wait with the network communication until Sunday or even Monday, and by that avoiding the problem which creates ghosts in the first place. It sounds logical to me, but I personally don't want to take the risk of not getting enough work to last through the server downtime.

The main reason to cause us to drain caches to detach/reattach to the project is for the Ghosts to be resent at a higher priority. We also need to work out the resends faster as the servers are running out of WU storage space. You want new work we need to get rid of/finish off the old stuff. See todays tech note.

---

If you check those new pendings you will see a lot of them will clear quickly as soon as the validater catches up. As for the new server,the bottleneck is in the bandwidth. The new server will help with storage and reliability but can't do anything about the bandwidth.

Couldn't the servers down throttle the download speed of each host when it's maxed out?

If you check those new pendings you will see a lot of them will clear quickly as soon as the validater catches up. As for the new server,the bottleneck is in the bandwidth. The new server will help with storage and reliability but can't do anything about the bandwidth.

Couldn't the servers down throttle the download speed of each host when it's maxed out?

As I understand it, the problem is not with download speed, but total bandwidth and communications between the Scheduling Process and the download servers, or the download servers and your hosts.

Ghosts are generated when the Scheduler assigns Tasks to your computer, but due to total demand on the servers, either the download order doesn't get to the Download servers, or the Download never gets to your computer, due to timeouts or dropped connections.

One of the things that WE can do to minimize this problem is for some of us to hold off on requesting new work during the first few hours immediately before or after the outage. This reduces the total number of requests the servers have to handle, making it less likely that a download will get "lost" in the chaos.

---

Donald
Infernal Optimist / Submariner, retired

Thanks folks.
I got them all done before the expiry, so all is well. Now they just have to get uploaded without them getting eaten by all the undead that must be lurking out there in DBland one the uploads start again tomorrow.

Len

Pending credit: 707,919.39

:D

---

Rig name: "x6Crunchy"
OS: Win 7 x64
MB: Asus M4N98TD EVO
CPU: AMD X6 1055T 2.8(1,2v)
GPU: 2x Asus GTX560ti
Member of: Dutch Power Cows