|
Bugzilla – Full Text Bug Listing |
| Summary: | multiple ntfsresize | ||
|---|---|---|---|
| Product: | [openSUSE] SUSE LINUX 10.0 | Reporter: | Joachim Reichelt <Joachim.Reichelt> |
| Component: | Installation | Assignee: | Thomas Fehr <fehr> |
| Status: | RESOLVED FIXED | QA Contact: | Klaus Kämpf <kkaempf> |
| Severity: | Normal | ||
| Priority: | P5 - None | CC: | szaka, yast2-maintainers |
| Version: | Beta 3 | ||
| Target Milestone: | --- | ||
| Hardware: | Other | ||
| OS: | All | ||
| Whiteboard: | |||
| Found By: | Other | Services Priority: | |
| Business Priority: | Blocker: | --- | |
| Marketing QA Status: | --- | IT Deployment: | --- |
| Attachments: | YaST2 log of installation | ||
|
Description
Joachim Reichelt
2005-08-29 06:25:34 UTC
Please provide YaST2 logs of installation, see: http://www.opensuse.org/index.php/Bug_Reporting_FAQ#YaST I can't comment why changes cause running ntfsresize all the time but I can tell why ntfsresize runs for 2 minutes. The performance is always dominated by the disk speed and since ntfsresize always checks NTFS consistency thus it reads the metadata of all files which are 1 KB. In average people have 100,000-200,000 files thus 100-200 MB needs to be read from the disk. Having a 20 MB/sec disk, this takes 5-10 seconds. If the disk DMA isn't setup properly then the speed could be only 1-2 MB/sec which means the task will take 50-200 seconds. Also, if you have 1 million files then the process will take 50 seconds (2 million files => 100 seconds, etc). One way to get to know the number of files is running time ntfsclone --metadata --output metadata.img <ntfs-partition> and check the "Num of used MFT records" at the end of the output. hdparm -tT <device> will tell the raw speed of the disk. Created attachment 48260 [details]
YaST2 log of installation
I did a time ntfsresize ...
I know, that there are lots of files, so I want only ONE, or better NO run of
ntfsresize, as long as these partitons are not affected. Isn't it better to run
ntfsresize only if I want/have to access these disks?
If there is space for SUSE, do not run it at all.
The current ntfsresize calls during installation is done since YaST2 needs to know if a ntfs filesystem is a possible candidate for resizing, this data is needed to make the proposal, so these calls are there for a reason. I was not aware that these calls need such a long time under some circumstances. I will see how easy it is to improve caching of this information. This is certainly possible, but I am not sure if the needed changes are not too invasive to do at this stage of development for SL 10.0. Added caching of this information to libstorage so that is is at least not determined multiple times. |