Bug 1452357
| Summary: | realtime-virtual-guest: not safe to run PMD thread with fifo:1 | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 7 | Reporter: | Luiz Capitulino <lcapitulino> |
| Component: | tuned | Assignee: | Jaroslav Škarvada <jskarvad> |
| Status: | CLOSED ERRATA | QA Contact: | Tereza Cerna <tcerna> |
| Severity: | unspecified | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | 7.4 | CC: | atragler, hhuang, jeder, jskarvad, juzhang, mtessun, mtosatti, ovasik, pezhang, sgordon, tcerna, thozza |
| Target Milestone: | rc | ||
| Target Release: | --- | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | tuned-2.8.0-5.el7 | Doc Type: | If docs needed, set a value |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2017-08-01 12:35:21 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | |||
| Bug Blocks: | 1240765, 1438120 | ||
Upstream commit: https://github.com/redhat-performance/tuned/commit/1ab378cc4a4372ca031bb765f656ab463b28af88 I've verified the kernel threads priorities in the guest are now correct. I've also run a quick cyclictest measurement test and it worked fine. Thank you, Luiz, for your test.
I did some other test. It works as expected -> VERIFIED.
Existing test was repaired for this issue. See /CoreOS/tuned/Regression/create-new-nfv-profiles.
===============================================
Verified in:
tuned-2.8.0-5.el7.noarch
tuned-profiles-nfv-guest-2.8.0-5.el7.noarch
PASS
===============================================
# tuned-adm profile realtime-virtual-guest
# ps axo pid,class,rtprio,comm | grep 'ktimersoft\|softirq\|rcuc\|rcub'
3 FF 2 ksoftirqd/0
12 FF 2 ksoftirqd/1
18 FF 2 ksoftirqd/2
24 FF 2 ksoftirqd/3
30 FF 2 ksoftirqd/4
34 FF 2 ksoftirqd/5
38 FF 2 ksoftirqd/6
42 FF 2 ksoftirqd/7
9144 FF 3 ktimersoftd
9150 FF 4 rcub
9156 FF 4 rcuc
===============================================
Reproduced in:
tuned-2.8.0-3.el7.noarch
tuned-profiles-nfv-guest-2.8.0-3.el7.noarch
FAIL
===============================================
# tuned-adm profile realtime-virtual-guest
# ps axo pid,class,rtprio,comm | grep 'ktimersoft\|softirq\|rcuc\|rcub'
3 TS - ksoftirqd/0
12 TS - ksoftirqd/1
18 TS - ksoftirqd/2
24 TS - ksoftirqd/3
30 TS - ksoftirqd/4
34 TS - ksoftirqd/5
38 TS - ksoftirqd/6
42 TS - ksoftirqd/7
9943 TS - ktimersoftd
9947 TS - rcub
9949 TS - rcuc
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2017:2102 tuned-2.9.0-0.1.rc1.fc26 has been submitted as an update to Fedora 26. https://bodhi.fedoraproject.org/updates/FEDORA-2017-d9c6b990df tuned-2.9.0-1.fc27 has been submitted as an update to Fedora 27. https://bodhi.fedoraproject.org/updates/FEDORA-2017-0e45ce4685 tuned-2.9.0-1.fc26 has been submitted as an update to Fedora 26. https://bodhi.fedoraproject.org/updates/FEDORA-2017-c30e9bd1ea |
Description of problem: The realtime-virtual-guest profile doesn't allow for running a PMD thread with fifo:1 priority in the guest. This is so because the profile doesn't organizes CPU bound kernel threads into FIFO slots as done by the realtime-virtual-host profile. This is how the realtime-virtual-host profile organizes kernel threads into FIFO slots: 151 FF 4 [rcuc/15] * 153 FF 3 [ktimersoftd/15] * 154 FF 2 [ksoftirqd/15] * As we can see, rcuc > ktimersoftd > ksoftirqd. This allows for running a PMD thread with fifo:1 without running into the risk of starving those important threads. However, the realtime-virtual-guest doesn't do any of this: 17 FF 2 [rcuc/1] * 19 FF 1 [ktimersoftd/1] * 20 TS - [ksoftirqd/1] * So, a PMD thread running in the guest with fifo:1 can starve ktimersoftd and ksoftirqd should those threads ever become runnable. The solution should be the realtime-virtual-guest profile to assign FIFO priority to kernel threads the same way the realtime-virtual-host is doing. Version-Release number of selected component (if applicable): tuned-2.8.0-3.el7.noarch