Bug 184469
Summary: | irqbalance and cpuspeed daemons cannot start | ||
---|---|---|---|
Product: | Red Hat Enterprise Linux 4 | Reporter: | Niksa Jurinovic <niksa> |
Component: | kernel-utils | Assignee: | Neil Horman <nhorman> |
Status: | CLOSED ERRATA | QA Contact: | Brian Brock <bbrock> |
Severity: | medium | Docs Contact: | |
Priority: | medium | ||
Version: | 4.0 | CC: | jbarton, lizhang, rbiba, travnicj-priv |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | i386 | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | RHBA-2007-0282 | Doc Type: | Bug Fix |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2007-05-07 23:34:01 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: | |||
Bug Depends On: | |||
Bug Blocks: | 227845 |
Description
Niksa Jurinovic
2006-03-08 23:26:33 UTC
Same problem here with irqbalance and cpuspeed, though I didn't try to compile anything. Running cpuspeed from the command line: server# cpuspeed Error: Could not open file for writing: /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor Error: No such file or directory server# There is no directory "cpufreq" inside cpu0, and I cannot create one: server# mkdir cpufreq mkdir: cannot create directory `cpufreq': Operation not permitted The fstab entry looks like this: none /sys sysfs defaults 0 0 I have the cpuspeed problem on 2 RHEL4 systems, and the irqbalance problem on 1 RHEL4 system. The system with both problems is a DELL Poweredge 600SC, and the system with just the cpuspeed problem is a DELL Poweredge 750 rack-server. "cat /proc/cpuinfo" shows both systems are running at their maximum speed. For what its worth, I have both problems on an FC5 box (Athlon XP1800+). Using (on the RHEL4 boxes): kernel-2.6.9-34.0.1.EL kernel-utils-2.4-13.1.80 Well, don't know about `cpuspeed', it didn't complain any errors so I haven't checked it. But I think I know what's going on with `irqbalance'. Too bad the CentOS Vault server is KO at the moment, so I can't check the previous RPM. I did only install the latest `kernel-utils' today and soon noticed the problem. It seems the problems with `irqbalance' only appears with the uniprocessor kernel (package `kernel'). Everything works just right with the SMP one (package `kernel-smp'). I don't use the `highmem' variant, so can't speak for that one. Obviously `irqbalance' is not intended for the uniprocessor kernel, so it is quite logical it fails (but not really pretty) in this case. Apparently the source of problem is as follows: From irqbalance.c, lines 68--69, function main(): if (cpucount < 2). exit(EXIT_SUCCESS); /* UP balancing is useless */ This causes the `/usr/sbin/irqbalance' to terminate without the calling script, `/etc/init.d/irqbalance' even knowing the daemon actually didn't start. The script then happily creates the lock file and reports success. Upon system shutdown, since the subsys is locked (ie. lock file exists) the rc script attempts to stop the service, which will report a failure due to `killproc irqbalance' command expectable failure. If I understand and recall correctly, there was some behaviour change in this regard recently. I have a feeling the `irqbalance' was failing at the startup before. The bug #107122, comment #2 suggests this (as also do some fading records in my memory...). Again, with SMP kernel everything works just fine (at least for me). The question, however is, if the current behaviour is correct? If success is reported and the service doesn't really start any daemon on uniprocessor kernel, it should also be OK no daemon is killed when the service is stopped. I can also quite confirm this problem didn't occur (or at least not in the form described) with uodate 2. I have in fact CentOS 4.2 installed which is currently being gradually updated to 4.3. These should correspond to RHEL 4's updates 2 and 3 respectively. This request was evaluated by Red Hat Product Management for inclusion in a Red Hat Enterprise Linux maintenance release. Product Management has requested further review of this request by Red Hat Engineering, for potential inclusion in a Red Hat Enterprise Linux Update release for currently deployed products. This request is not yet committed for inclusion in an Update release. I fixed this by adding some logic to the irqbalance init script. I didn't want to break any cases in which silent failure for the U.P. case was expected when irqbalance is used in another script (several people seem to do this for custom irq balancing purposes). This maintains the current behavior for the application binary, but fails the service and logs an error in /var/log/messages in the event that "service irqbalance start" is run on a uniprocessor system. An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on the solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHBA-2007-0282.html |