OpenSolaris

Printable Version Enter a New Search
Bug ID 6573681
Synopsis deleting a very large file can be slow
State 10-Fix Delivered (Fix available in build)
Category:Subcategory kernel:zfs
Keywords rtiq_reviewed
Responsible Engineer Mark Maybee
Reported Against s10 , alpha , s10_45 , snv_65 , s10u5_fcs , solaris_10u5
Duplicate Of
Introduced In solaris_nevada
Commit to Fix snv_94
Fixed In snv_94
Release Fixed solaris_nevada(snv_94) , solaris_10u7(s10u7_01) (Bug ID:2167881)
Related Bugs 6600154 , 6606810 , 6721941 , 6723124 , 6724798 , 6728399 , 6730750 , 6759570 , 4940368
Submit Date 25-June-2007
Last Update Date 20-October-2008
Description 
when deleting a very large file, we won't be able to 
cache all the level-1 indirect blocks from when we 
prefetch them in dmu_tx_count_free() until when we 
process the free in syncing context.  So, free_children() 
should kick off multiple i/os in parallel (say, all of 
the l1's under each l2).
Work Around 
N/A
Comments 
N/A