|
Bugzilla – Full Text Bug Listing |
| Summary: | package manager is very very slow | ||
|---|---|---|---|
| Product: | [openSUSE] SUSE Linux 10.1 | Reporter: | Jon Nelson <jnelson-suse> |
| Component: | Basesystem | Assignee: | Klaus Kämpf <kkaempf> |
| Status: | RESOLVED FIXED | QA Contact: | E-mail List <qa-bugs> |
| Severity: | Normal | ||
| Priority: | P5 - None | ||
| Version: | Beta 5 | ||
| Target Milestone: | --- | ||
| Hardware: | i686 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Found By: | Other | Services Priority: | |
| Business Priority: | Blocker: | --- | |
| Marketing QA Status: | --- | IT Deployment: | --- |
|
Description
Jon Nelson
2006-02-26 01:44:34 UTC
OK this is not connected to the new packager backend as it was introduced in b4. Since b4 however, there is a lot of logging activity to make debugging easier with the new backend. This could be a problem with RPM itself. Try executing `time rpm -e kernel-source', how long does this take? Couldn't find kernel-source, I used OpenOffice_org (monster rpm!) 19s to remove 1m30s to rpm -Uvh That's rather less time than I saw the installer take for much, much smaller rpms. I could be mistaken about beta3, but I don't think so. I could try a reinstall in the next few days, but it takes /time/ and I don't have much to spare. OK this seems to be another problem then. At first correct the hardware-field (which is `other' atm) and attach the logs in /var/log/YaST2. I'd like to note that general slowness of installation has been a problem for as long as I can remember using SUSE, however with the recent betas that slowness has been magnified considerably. Please attach the YaST logfiles, I'll reassign the bug after that. New development: I think the team around zypp is currently working on this problem and fixing some performance issues. I'll reassign this to Klaus for a comment. Several things lead to this - insufficient information about language dependencies, so the resolver has to evaluate multiple branches (fixed in beta6) - excessive debug (well, we're still in development) - std::set in libstdc++ is very slow (under investigation) Cool - beta6 was much faster, although still not up to par. I can only assume that plenty of debugging code and other known factors are contributing, so please just take this as a "beta 6 is much better" comment. Well, 'up to par' depends on the scale you're applying ;-) It could be made fast by adding more hashes, which in turn will increase memory consumption. However, it will probably not be as fast as YaST in 10.0 and before because the new solver calculates more scenarios. Anyway, I consider this fixed. |