Bug 155591 - [zypp] transact (zmd backend) is too slow
Summary: [zypp] transact (zmd backend) is too slow
Status: VERIFIED FIXED
: 171457 (view as bug list)
Alias: None
Product: SUSE Linux 10.1
Classification: openSUSE
Component: libzypp (show other bugs)
Version: Beta 6
Hardware: Other Other
: P5 - None : Normal (vote)
Target Milestone: ---
Assignee: Klaus Kämpf
QA Contact: Nat Budin
URL:
Whiteboard:
Keywords:
Depends on:
Blocks: 176301
  Show dependency treegraph
 
Reported: 2006-03-06 21:55 UTC by Klaus Kämpf
Modified: 2006-06-16 21:50 UTC (History)
3 users (show)

See Also:
Found By: Other
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Klaus Kämpf 2006-03-06 21:55:13 UTC
It build the complete repository although resolve-dependencies ran before and already calculated the needed ones.
So transact just has to look at those packages mentioned in the
transactions table.
Comment 1 Wade Berrier 2006-03-25 21:03:07 UTC
In general, the native processes (as in, outside of mono) that get run require huge amounts of cpu processing (parse, query, transact).

I'm a long time debian/apt user, so I'm _really_ _really_ excited about the zmd integration.  I'm also hoping that zmd will be able to run factory in a practical manner on a 500 Mhz type machine.  (ex: apt drives Debian's huge repo wonderfully on a low end machine)

I realize that apt isn't using sqlite nor xml.  But in order to be well accepted, I think it's really important for this stuff to be fast (installing packages from the command line, updating all packages, etc...).
Comment 2 Joachim Werner 2006-04-23 15:05:31 UTC
Klaus, can we assign this bug to anybody else to get it fixed, or does it have to be on your plate? From the description I don't see if this is a problem in Mono, and more generally what component exactly causes the problem.
Comment 4 Klaus Kämpf 2006-05-02 10:12:26 UTC
*** Bug 171457 has been marked as a duplicate of this bug. ***
Comment 8 Naresh Wignarajah 2006-05-15 16:31:29 UTC
Assigning to Nat to test and provide test data.  CC'ing Thorsten.
Severity -> Critical based on user feedback on slowness.
Comment 9 Naresh Wignarajah 2006-05-16 15:30:43 UTC
Oops, corrected Severity.
Comment 13 Klaus Kämpf 2006-06-14 15:03:20 UTC
Optimized versions of helpers have been submitted to autobuild.
Comment 14 Nat Budin 2006-06-15 21:43:59 UTC
Re comment #13: Klaus, AFAICT, yesterday's helper optimizations only affected parse-metadata and service-delete, so that shouldn't affect this bug...?
Comment 15 Klaus Kämpf 2006-06-16 07:39:07 UTC
All helpers are performing better now, including 'transact'.
Comment 16 Nat Budin 2006-06-16 21:50:35 UTC
Yep, I re-ran it with ltrace, and the addResolvables call, although still (slightly) over 50% of the overall time, is significantly less than before.  Thanks, Klaus!

(We should probably revisit this for SP1 to see about making it Even Faster...)