If you want to see "bunkering" in action, take a look at the hosts in this list. Several hours ago, these hosts had zillions of tasks on board but virtually nothing returned. Now the results seem to be flooding in.
What is also interesting is that there are multiple hosts with identical hardware all with close hostIDs. Perhaps they are different physical machines but I'm wondering if (in order to get around daily limits) they are the same machine with multiple IDs. No wonder there has been quite a distortion to normal operations.
There are several ways to "bunker" tasks and running multiple clients with different data directories is the most prominent I guess. There are several threads in team forums dedicated to explaining how this is done. Another way is to setup several Linux VM which also get different hostids to circumvent task limits.
This is my first Pentathlon as a project administrator, the last two years I only participated as a user of my team. Everyone here is very confident that our server can handle the validation of several days worth of tasks within a single day and after that it will be back to normal. It will be interesting to look at some graphs after the Pentathlon to see how much the actual throughput was affected.
OCN has been bunkering since it was announced as a CC project in the Pentathlon. I've made a guide on how to setup multiple clients and people have been using it to get a lot of work. The calculated ETA is way off to start on a fresh client so several clients had to be setup on the same machine to keep the GPUs fed. 4h10min vs actual run time of 11-17min. Some people there had 5 digit count of tasks waiting to upload after the event start. That does affect everyone though but it should even back out after the event. I'm sure there will be some dumped WUs that people won't want to crunch.
I'm glad E@H has been able to handle the work. This is my 2nd Pent and E@H seemed to have handled it the best of any so far.
When the pentathlon ended the large increase in pendings stopped and I expected to see them eventually revert to previous levels. That is not what has happened. What has happened is they remain at about double of what they were and continue to climb slowly. This seems to be a new normal.
If you want to see
)
If you want to see "bunkering" in action, take a look at the hosts in this list. Several hours ago, these hosts had zillions of tasks on board but virtually nothing returned. Now the results seem to be flooding in.
What is also interesting is that there are multiple hosts with identical hardware all with close hostIDs. Perhaps they are different physical machines but I'm wondering if (in order to get around daily limits) they are the same machine with multiple IDs. No wonder there has been quite a distortion to normal operations.
Cheers,
Gary.
There are several ways to
)
There are several ways to "bunker" tasks and running multiple clients with different data directories is the most prominent I guess. There are several threads in team forums dedicated to explaining how this is done. Another way is to setup several Linux VM which also get different hostids to circumvent task limits.
This is my first Pentathlon as a project administrator, the last two years I only participated as a user of my team. Everyone here is very confident that our server can handle the validation of several days worth of tasks within a single day and after that it will be back to normal. It will be interesting to look at some graphs after the Pentathlon to see how much the actual throughput was affected.
OCN has been bunkering since
)
OCN has been bunkering since it was announced as a CC project in the Pentathlon. I've made a guide on how to setup multiple clients and people have been using it to get a lot of work. The calculated ETA is way off to start on a fresh client so several clients had to be setup on the same machine to keep the GPUs fed. 4h10min vs actual run time of 11-17min. Some people there had 5 digit count of tasks waiting to upload after the event start. That does affect everyone though but it should even back out after the event. I'm sure there will be some dumped WUs that people won't want to crunch.
I'm glad E@H has been able to handle the work. This is my 2nd Pent and E@H seemed to have handled it the best of any so far.
Disrupting projects for your
)
Disrupting projects for your amusement seems very rude to me.
When the pentathlon ended the
)
When the pentathlon ended the large increase in pendings stopped and I expected to see them eventually revert to previous levels. That is not what has happened. What has happened is they remain at about double of what they were and continue to climb slowly. This seems to be a new normal.