Alexander Barkov skrev:
> 

>                  Name: cache.c.gz


Splitting with the new cache.c. First run:

-rw-r--r--    1 root     root         4664 Feb 15 19:38 982006130.del 
-rw-r--r--    1 root     root      2824548 Feb 15 19:38 982006130.wrd 
-rw-r--r--    1 root     root        22800 Feb 15 19:38 982019478.del 
-rw-r--r--    1 root     root      8216488 Feb 15 19:38 982019478.wrd 
-rw-r--r--    1 root     root          832 Feb 15 19:38 982024249.del 
-rw-r--r--    1 root     root       151860 Feb 15 19:38 982024249.wrd 
-rw-r--r--    1 root     root           88 Feb 15 19:38 982024900.del 
-rw-r--r--    1 root     root        42176 Feb 15 19:38 982024900.wrd 
-rw-r--r--    1 root     root        54360 Feb 15 19:38 982027284.del 
-rw-r--r--    1 root     root     31465324 Feb 15 19:38 982027284.wrd 
-rw-r--r--    1 root     root        14240 Feb 15 19:38 982027618.del 
-rw-r--r--    1 root     root      8815804 Feb 15 19:38 982027618.wrd 

completed without any problems.

Second run (note the sizes):

-rw-r--r--    1 root     root        59920 Feb 15 19:38 982040748.del
-rw-r--r--    1 root     root     31457740 Feb 15 19:39 982040748.wrd
-rw-r--r--    1 root     root         1480 Feb 15 19:39 982040807.del
-rw-r--r--    1 root     root       637240 Feb 15 19:39 982040807.wrd
-rw-r--r--    1 root     root        51920 Feb 15 19:39 982045300.del
-rw-r--r--    1 root     root     31469304 Feb 15 19:39 982045300.wrd
-rw-r--r--    1 root     root        69248 Feb 15 19:39 982047843.del
-rw-r--r--    1 root     root     30213344 Feb 15 19:39 982047843.wrd
-rw-r--r--    1 root     root        10232 Feb 15 19:39 982099518.del
-rw-r--r--    1 root     root      6195408 Feb 15 19:39 982099518.wrd

completed without any problems. The number of words found in 
the index increased considerably compared with the previous 
tree ("and": 24000/25614,[*] "something": 8635/8999, "nothing": 
7321/7665, "all": 18119/18889), which means that every crash 
took with it part of the index too. 

Also, I have the impression that these splittings went much faster 
than previous ones. What I found a bit strange is that there was no 

  Deleting from tree/nn/n/nnnnnnnn 

activity at all. The splitting output looked like this *only*:

/var/mn3110-3/tree/FF/F/FFFE8000 old:   8 new:  12 total:  20
/var/mn3110-3/tree/FF/F/FFFE9000 old:   0 new:   1 total:   1
/var/mn3110-3/tree/FF/F/FFFEA000 old:   1 new:   2 total:   3
/var/mn3110-3/tree/FF/F/FFFEC000 old:   1 new:   1 total:   2
/var/mn3110-3/tree/FF/F/FFFF6000 old:  12 new:  15 total:  27
/var/mn3110-3/tree/FF/F/FFFF8000 old:   1 new:   1 total:   2
/var/mn3110-3/tree/FF/F/FFFFE000 old:   0 new:   2 total:   2

So when does splitter delete from tree files? 

Z


[*] Yes, I know, I am intentionally not using stopwords. 



-- 
oracle@everywhere: The ephemeral source of the eternal truth...
___________________________________________
If you want to unsubscribe send "unsubscribe general"
to [EMAIL PROTECTED]

Reply via email to