[SCore-users] (no subject)

[neural_shock at e-mail.ru]

hi, it is me again, mike. i have two questions.

1st. if there some ability to use SCore's checkpointing with other cluster schedulers,
for example with pbs, with pure pbs, i mean, without tss and groups?

2st. i have put to the trial checkpointing mechanism. here is situation i liked
it to handle. i have two computing hosts ( PIII 800MHz/128Mb ), say comp1 and
comp2. comp1 is scored server in group pcc and it is beeing rebooted by cron
every ten minutes to simulate system failure. on these hosts LU class A benchmark
from MPB is running. 

and i have observed the following behavior of system. when comp1 node starts
the rebooting in the middle of checkpointing process some kernel error takes
place, according to eip and System.map in function __free_pages_ok. several
times this error does not lead to "system crash" ( node reboots normally and
computation restarts from normally finished checkpoint ), but later node cannot
complete the "going down" process( not linux's process, i use this term to name
actions which system carries out before rebooting itself ), and halts.

may be it is some kind of bug, may be i have done something wrong. and if somebody
wish to look at details i can post debug messages log.

with respect, mike.
http://www.e-mail.ru

---
������� ����� �� DVD �� ���� �� 250�. �������, ����������� ����������.
���������� DVD ������ �� http://www.auction.ru/v2/Catalogue/Catalogue.asp?RID=73