This seems very similar to bug#64836 reported against RHEL3. As far as I can tell, this just popped up in RHEL4 after update 2 (I updated just last week and this Sunday had a bunch of these messages). Anyone else confirm that they didn't have it before, and now is happening?
Yeah, after we updated, we started getting these messages from our machine running RHEL4 AS. I don't get them on any of my machines running RHEL4 WS, but they have different packages installed too.
Hello, test version of man which contains patch against this bug is available: http://people.redhat.com/varekova/man-1.5o1-9.test.src.rpm and http://people.redhat.com/varekova/man-1.5o1-9.test.i386.rpm. Could you please test it?
I'd be most interested in the results of the patch as to wether it resolves the problem before I go and install it on a client's production system. I've been having this problem for a number of months now.
Created attachment 122683 [details] patch used in man-1.5o1-9.test
Isn't this patch just throwing error messages away? Wouldn't it be better to actually not have the errors in the first place?
Shiraz, you are right. The patch is only erasing "broken pipe" error messages, the real problem is in other package. It will be better to fix the reason of this error messages, but this patch does not exist now so and man patch fixes this problem.
cron.daily/00-makewhatis: weekday=`/bin/date +%u` if [ "$weekday" = "7" ] then makewhatis -w else makewhatis -u -w fi Then we can throw away cron.weekly/makewhatis and the lockfiles as well. The logic of keeping weekly jobs in cron.weekly is defeated anyway, either by the lockfile or by the bug.
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on the solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHBA-2006-0344.html
The package mentioned in the above errata is dated 2004. How can a fix from 2006 already exist in this package? We're still seeing the makewhatis problem using man-1.5o1-9 which is the only version of man available in the RHEL 4 RHN channels.
Same thing still happening here, too.
This appears to have come back as of the makewhatis run of last night (16 July 2006). Same symptoms as before. Looks like the patch got zapped.
> Looks like the patch got zapped. I would rather assume that recent updates to vixie-cron made a workaround ineffective.
Looking closer it appears that 'man-1.5p-pipe_makewhatis.patch', i.e. --- man-1.5p/src/makewhatis.sh.pipe 2005-05-12 15:23:07.635109152 +0200 +++ man-1.5p/src/makewhatis.sh 2005-05-12 15:24:19.780141440 +0200 @@ -208,9 +208,9 @@ sub(/\..*/, "", progname); if (use_zcat || use_bzcat) { if (use_zcat) { - pipe_cmd = "zcat \"" filename "\""; + pipe_cmd = "zcat \"" filename "\" 2>/dev/null"; } else { - pipe_cmd = "bzcat \"" filename "\""; + pipe_cmd = "bzcat \"" filename "\" 2>/dev/null"; } # try to avoid suspicious stuff if (filename ~ /[;&|`$(]/) { was never applied in an RHEL context and vixie-cron updates from the last week brought that to light.
*** Bug 199203 has been marked as a duplicate of this bug. ***
This problem is fixed in man-1.5o1-9.rhel4. man-1.5o1-9 is affected with this bug. Please ask GSS about man-1.5o1-9.rhel4 package. If there is anybody who has a system with man-1.5o1-9.rhel4 which is affected by this bug please add a comment here.
Please see comment #17 which I submitted. We are seeing the problem with man-1.5o1-9.rhel4 because the problem that was fixed in RHEL 3 was never ported to RHEL 4.
>> If there is anybody who has a system with man-1.5o1-9.rhel4 which is affected by this bug please add a comment here. << RHEL4 + man-1.5o1-9.rhel4 = still see trouble
Created attachment 132900 [details] Test version of makewhatis script. Could you please attach your actual makewhatis script (/usr/bin/makewhatis). And could you please replace the actual version of makewhatis script by attached test version. (The output of attached test version is more verbose) and attach the weekly output of this version to bugzilla comment.
I have here RedHat Enterprise Linux AS 4, and I see the same errors since the last vixie-cron update. However the fixed man package is not available in the rhel-x86_64-as-4 channel of RedHat Network. Could this fixed package be made available there as well?
Created attachment 132914 [details] /usr/bin/makewhatis from RHEL 4 Please re-open this bug as it's clear we still have a bug in RHEL 4.
Please Brian could you attach the version (release number) of your man - package?
man-1.5o1-9. In comment you mention man-1.5o1-9.rhel4 which does not exist to my knowledge.
Created attachment 132919 [details] Verbose output from /usr/bin/makewhatis test script This 50+MB of output is from your test script.
Can I get someone to test this in a test environment. I have been working on porting the bug fix forward to RHEL4. Can I get you to test the rpm packages available from: http://people.redhat.com/wmealing/RPMS/RHEL4/man/ These are test packages only, and should not be used on a production system. Red Hat GSS will not support these packages. Assuming it does solve the issue I'll attempt to get the fix in the next release.
Moved this bug back to 3, as it originally was. RHEL 4 bug is at https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=201817
*** Bug 201818 has been marked as a duplicate of this bug. ***
The rhel3 version of this bug is 64836, this bug is against rhel4.
*** Bug 201817 has been marked as a duplicate of this bug. ***
*** Bug 201526 has been marked as a duplicate of this bug. ***
What is the status of getting an errata for this? Any chance it could be considered for Fastrack?
The link to errata for this problem is in comment #15.
No, that errata for RHEL 4 does not fix the problem which is why this bug is supposed to be re-open. Only the RHEL 3 man errata fixed the problem for RHEL 3. There has been no RHEL 4 errata which fixes this issue. I still have an open Issue Tracker 94958 in regards to this issue and I was given a test package "man-1.5o1-10_TEST.x86_64.rpm" which DOES fix the problem. I'll follow up in IT as well.
Brian I look at comment #28 there is your makewhatis patch - which is not fixed, but I just grub man-1.5o1-9.rhel4.i386 and it's version of makewhatis script contains patch against this bug. Please could you try re-install your man page package and send me /usr/sbin/makewhatis script and output of "rpm -q man" command again. Thanks.
From one of my (knowledgeable) users: Version of man pages has nothing to do with it. Somehow yum or rpm environment is set up to ignore sigclose and when post script restarts crond during update, crond inherits those environment settings. Restarting crond by hand fixes this problem. To observe this, update crond and: grep SigIgn /proc/{cron pid}/status then restart crond and compare. Minor, easy to fix problem after all.
I copy/pasted comment 42 into a new bug (475106) in hopes of getting someone's attention to this 3-year-old issue. (We can't reopen this bug.)