On Friday 26 January 2007 17:37, Craig White wrote: >On Fri, 2007-01-26 at 15:59 -0500, Gene Heskett wrote: >> OTOH this FC6 system has been strange from the gitgo, like having to >> go find a bunch of cron related files from the old FC2 install and >> copy them over before cron would run. None of its config and >> reference files were installed by anaconda. They simply weren't >> there. And all the rpm checking I could get it to do said the install >> was all right. Yeah, sure. > >---- >which came first, the chicken or the egg? > A very good question. >Little doubt in my mind that copying the files over from an FC-2 box to >a new FC-6 install would create a bunch of problems with selinux which >of course you can't solve because the man pages aren't up to your >standards. While you ***may*** have solved the problems by disabling >selinux (I say may because as of yesterday, you were still uncertain >whether you had actually disabled selinux), I am reasonably convinced by >your recollections of what you have done to your install that most of >these problems are self inflicted. Nope, in this case by anaconda. I did NOT delete the cron stuff, and in fact wasn't aware of it until I I tried to run kcron and add my scheduled stuff to the new install. No common crontab file=kcron crash. So I started comparing the cron stuff installed on /dev/hda with the stuff on /dev/hdb, which was, and still is, my old FC2 system disk. Everytime I found something on hdb that wasn't on hda, it got copied. Just text files, all of them. Cron itself would start, but 5 seconds later a service crond status said it wasn't running. >To that end, the fonts in /root - they didn't get there from installing >the msttcorefonts package because even if you build the rpm as root, >they would be put into /usr/src/redhat/SOURCES, and when you install the >rpm (obviously as root), they go into /usr/share/fonts/msttcorefonts >directory and the only way for them to get into /root is to manually >copy them there...and by your post yesterday, the security contexts >would have changed exactly as they had by doing something similar to... >cp -r /usr/share/fonts/msttcorefonts /root A, they never were in /root/.fonts even on the FC2 install. B, they are in /usr/share/fonts/msttcorefonts, on both the FC6 install on hda, and on the FC2 install thats now mounted from hdb. C, there are some fonts in the old FC2 disks /root/.fonts dir, and those were copied to hda's /root/.fonts for with mc, which probably duplicates the cp performance. D, I'd run fixfiles a week, maybe 10 days ago, which took several hours because it walked through every disk mounted instead of staying on its filesystem. That's a bug to me, but a shrug ATM. E, It had all worked, including pulling in some fonts from extras with yumex quite early after I put FC6 on this box. I had also used mc to copy over a dozen or more fonts I'd downloaded from goldenweb.it, putting them in /usr/share/fonts where they were instantly available to OOo with no intervention on my part. F, printing dies 3+ days ago after putting in a ready made rpm of msttcorefonts-2.x.noarch. I've taken that rpm apart and looked at 99% of its bytes with various editors and AFAICT, that rpm is clean and legal. G, 2.5 days later, after I had deleted some font.dir and font-cache files of 0 length, printing starts working again. H, /var/log/secure contains nothing from selinux since the logrotation sunday morning. I, fc-cache is still broken. >It is clear that your pattern continues to take the easy path - >regardless of the implications for system safety, whether it is how you >log in, use the Desktop as root, build packages as root, install >directly from tarballs instead of building rpms, etc. and thus, it is of >little surprise that you claim a disproportionate amount of problems >related to packaging. The thing about packaging is the consistency...and >if there was a problem with the packaging, everyone who installed the >same packages (which is basically everyone since we are talking about >vixie cron), would have had the same problem - unless of course, the >packagers had some means to single you out at installation time. I have NDI why cron wasn't fully installed. I did not do anything to intervene in anaconda's feeble minded attempts to build me a working system. Humm, another ugly thought come to mind, that install was done from a dvd I burnt, from a torrent download, so the kernel bug that has existed for many versions that effects pre-allocated file writes on a random LSB offset, exactly the way a torrent operates, and which was finally fixed in 2.6.20-rc3(IIRC, maybe if was rc4) might have munged that dvd. But IIRC the checksum was good. But its got me thinking. sha1sum of FC-6-i386-DVD.iso: 6722f95b97e5118fa26bafa5b9f622cc7d49530c FC-6-i386-DVD.iso from SHA1SUM file: 6722f95b97e5118fa26bafa5b9f622cc7d49530c FC-6-i386-DVD.iso So that idea is well and truly shot down, and I always have k3b check the sum of the burned disk. Idea, maybe the crontab and supporting files are in a different package? Nice idea but rpm says its installed from crontabs.noarch which srcs in crontabs-1.10-8 And several dozen other packages according to a yum provides crontab listing. So I have NDI where the heck they went, all I know for sure is that they weren't there when I went looking to see why cron driven stuff wasn't working the next morning after the install of FC6. >Craig -- Cheers, Gene "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Yahoo.com and AOL/TW attorneys please note, additions to the above message by Gene Heskett are: Copyright 2007 by Maurice Eugene Heskett, all rights reserved.
