Re: problem with maxpromoteday set to 0
2007-10-09 08:29:31
Sorry, i forgot one thing: i've already forced on Friday all DLE to do a
full, but it was the same problem the week after.
Thomas Ginestet
Jean-Louis Martineau a écrit :
They are not due for a full dump, why do you think they should do one?
With your config, they will do a full at every 7 days.
These DLE are done Monday, the will continue to be done Monday, that's
your config.
You must force them to the next full on Friday with (amadmin <config>
force serv1).
Every time a DLE will not get it's full on Friday you will get the
same problem. It's better to force all DLE to do a full.
The Friday, before amdump, you should run: amadmin <config> force \*
Jean-Louis
Thomas Ginestet wrote:
Hi list,
I've set 0 to the maxpromoteday parameter but i've got 2 (sometimes
3) of my 53 DLE which promote the full dump one day after all the
others.
I have a dumpcycle of 7 days and a runspercycle of 5 (the full dump
is normally made on friday but for theses 2 DLE, it's made on monday).
No problem of space with the size of the tape.
I can find this in the amdump log:
pondering serv1:/work/org1... next_level0 3 last_level 2 (not due for
a full dump, picking
an incr level)
pick: size 3759920 level 2 days 2 (thresh 40960K, 1 days)
pick: next size 2308650... BUMPED
curr level 3 nsize 2308650 csize 1245984 total size 24078802
total_lev0 21969752 balanced-lev0size 15457024
Where the others DLE are due for level 0:
pondering serv1:/work/org2... next_level0 0 last_level 2 (due for
level 0) (picking inclevel for degraded mode) pick: size 79670
level 2 days 3 (thresh 40960K, 1 days)
pick: next size 29480... BUMPED
Any hint would be appreciated
Thomas Ginestet
|
|
|