[LU-1885] Lustre should be able to limit amount of memory used for read and write caches Created: 11/Sep/12  Updated: 16/Feb/19  Resolved: 16/Feb/19

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: New Feature Priority: Minor
Reporter: Shuichi Ihara (Inactive) Assignee: Peter Jones
Resolution: Won't Fix Votes: 0
Labels: None

Issue Links:
Related
is related to LU-10967 MDT page cache management improvements Open
Rank (Obsolete): 10112

 Description   

This has been discussed before, but I couldn't find any bugs or mailing list threads that specifically dealt with it, so I thought I'd open up a formal feature request. Basically we would like to use the read and write caches, but don't want them to take up all the memory. We'd like to be able to keep a large amount of memory available for the block bitmaps and other OST ext metadata, without having to pin everything in memory with vfs_cache_pressure=0.

How much control is there in the Linux page cache? Is it possible to tell it to start flushing the object caches if they start taking up X amount of space?



 Comments   
Comment by Kit Westneat (Inactive) [ 11/Sep/12 ]

or alternatively if it could prioritize it so the cache flushes objects before metadata, that would have the same desired effect.

Comment by Peter Jones [ 11/Sep/12 ]

thanks for the suggestion Kit. I'll raise this suggestion when we start the next round of planning

Comment by John Fuchs-Chesney (Inactive) [ 29/Jul/14 ]

Do we want to keep this ticket open?

Thanks,
~ jfc.

Comment by Peter Jones [ 16/Feb/19 ]

Closing out of date ticket

Generated at Sat Feb 10 01:20:32 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.