Bug 474492
Summary: | yumbackend.py likes to take all my CPU power | ||
---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Mark <markg85> |
Component: | PackageKit | Assignee: | Richard Hughes <richard> |
Status: | CLOSED NEXTRELEASE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> |
Severity: | medium | Docs Contact: | |
Priority: | low | ||
Version: | 10 | CC: | alreaud, ffesti, james.antill, katzj, lmacken, pmatilai, rhughes, richard, robin.norwood, tim.lauridsen, tuxbrewr |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2009-01-14 13:32:52 UTC | Type: | --- |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Mark
2008-12-04 01:24:03 UTC
Not really reassinging but asking for input. Richard, Tim, Does it make sense to nice-out the time that any of the pk backends get? I guess what the reporter is experiencing is either the auto-get-updates or maybe the auto-cache-refresh code, although the latter is unlikely as that only happens when idle for a while. I don't think it makes sense to kill a job just because it's taking a long time, although I do think it would be good to CPU limit this check if possible. I don't know how to do that in python. Richard. For more info (in case that wasn't clear) i'm using a fairly default fedora installation + the rpmfusion repo. i haven't changed any yum settings. The install was from the Fedora DVD (not the live CD) @richard, i have no idea which one of "auto-get-updates" and "auto-cache-refresh".. is there any yum log somewhere where i can check that? the yum.log file is not telling me anything about yumbackend.py yumbackend is run from PackageKit, to interface with yum. You can get all the details about what was done, why whom, and why, by doing this command "pkcon get-transactions" this produces a few screenfulls of: Transaction: /914_caebcbce_data System time: 2008-12-03T08:37:22.786753Z Succeeded: True Role: update-system Duration: 599 (seconds) Command line: gpk-update-viewer User ID: 500 Username: hughsie Real name: Richard Hughes Affected packages: - downloading samba-client-3.2.5-0.23.fc10.i386 - updating selinux-policy-3.5.13-26.fc10.noarch - updating selinux-policy-targeted-3.5.13-26.fc10.noarch ..... - updating samba-3.2.5-0.23.fc10.i386 - updating libsmbclient-3.2.5-0.23.fc10.i386 - cleanup samba-common - cleanup wine-ldap ..... - cleanup gtkhtml3 - cleanup qt-x11 I'm getting a strange output.. My last "transaction" was according to the line you gave me: [mark@localhost ~]$ pkcon get-transactions Transaction : /6_abbbdbeb_data timespec : 2008-12-02T01:28:59.262586Z succeeded : 1 role : install-packages duration : 0 (seconds) data : downloading tinyxml;2.5.3-3.fc9;x86_64;fedora A simple, small, C++ XML parser downloading wxBase;2.8.9-1.fc10;x86_64;fedora Non-GUI support classes from the wxWidgets library downloading filezilla;3.1.5.1-1.fc10;x86_64;updates FileZilla FTP, FTPS and SFTP client downloading wxGTK;2.8.9-1.fc10;x86_64;fedora GTK2 port of the wxWidgets GUI library installing wxBase;2.8.9-1.fc10;x86_64;fedora Non-GUI support classes from the wxWidgets library installing wxGTK;2.8.9-1.fc10;x86_64;fedora GTK2 port of the wxWidgets GUI library installing tinyxml;2.5.3-3.fc9;x86_64;fedora A simple, small, C++ XML parser installing filezilla;3.1.5.1-1.fc10;x86_64;updates FileZilla FTP, FTPS and SFTP client Which is not right. The transaction itself is fine but there should be more. i did a yum -y install updates yesterday which downloaded and installed for about 170 MB of updates but that's nowhere to be found in the transactions. All other transactions are from last month (november) commit 491c1c887abd42c282ee8277bc65456be616577b Author: Adel Gadllah <adel.gadllah> Date: Wed Dec 17 17:55:22 2008 +0100 Don't let spawned backends hog the disk by default :100644 100644 d94ce75... 56c8836... M etc/PackageKit.conf.in :100644 100644 601ed7d... 4553a21... M src/pk-spawn.c :000000 100644 0000000... c2b9e96... A src/pk-sysdep.c :000000 100644 0000000... c453e29... A src/pk-sysdep.h I've got an issue where I execute "yum update" after "yum check-update" and yum appears to hog all the computer resources, the hard disk starts thrashing, and the user is locked out of normal Gnome window functions such as scrolling, opening new applications, switching virtual terminals, etc. Upon examination with gnome-system-monitor, one sees that the kernel and user cpu use is minimal to moderate, in the 10%-30% range, but the IO waits are between 70% and 90%. This is using the panel display of gnome-system-monitor. There is no network activity at this time. The problem occurs regardless of whether you use yum from the command line or PackageKit from Gnome. The problem occurs regardless of whether you're installing or removing packages. Some data - Host: Windows XP SP3 VM: VMware 7 Client: Fedora 12 Kernel: vmlinuz-2.6.31.6-162.fc12.i686 yum: yum-3.2.25-1.fc12.noarch |