|Summary:||Yum does not check /var/cache/yum size|
|Product:||[Fedora] Fedora||Reporter:||Mikus Grinbergs <mikus>|
|Component:||yum||Assignee:||Seth Vidal <skvidal>|
|Status:||CLOSED UPSTREAM||QA Contact:||Fedora Extras Quality Assurance <extras-qa>|
|Version:||11||CC:||ffesti, james.antill, maxamillion, pmatilai, rc040203, sascha-web-bugzilla.redhat.com, tla|
|Fixed In Version:||Doc Type:||Bug Fix|
|Doc Text:||Story Points:||---|
|Last Closed:||2009-08-27 20:49:47 UTC||Type:||---|
|oVirt Team:||---||RHEL 7.3 requirements from Atomic Host:|
|Cloudforms Team:||---||Target Upstream Version:|
|Bug Depends On:|
Description Mikus Grinbergs 2009-08-21 20:12:30 UTC
Description of problem: A requested yum operation can suddenly run out of room in /var/cache/yum, causing yum to terminate abnormally. Version-Release number of selected component (if applicable): yum-3.2.23-3 How reproducible: Always, if /var/cache/yum too small. Steps to Reproduce: 1.Allocate say 100 MB to /var/cache/yum. 2.Request a yum operation that would require the downloading of more than 70 MB of packages. [Note: I had once before run into this problem when yum decided to install in my /var/cache/yum the "...-filelist.sqlite" data for the (F11) Fedora repository (almost 70 MB in size when unpacked). But now the problem came up with an ordinary 'yum upgrade', without even involving "...-filelist.sqlite".] Actual results: After downloading for quite a while, yum terminated with "out of space" messages. Expected results: Yum would have warned before starting the downloads that it did not have enough room in /var/cache/yum. Better yet, yum would have performed a "stepwise" download in case of "not enough room for all", fetching only a single package at a time (plus its dependencies) and procesing that by itself, instead of beginning by downloading *every* package identified by the requested yum operation. [What I had issued was 'yum upgrade', which found enough updateable packages to need more than 70 MB of downloads.] Additional info: This occurred on an F11 image running on an OLPC-XO1 machine (which has a very limited amount of real storage). Yet that image had mounted /var/cache/yum as a separate 'tmpfs' filesystem, with a total allocation of 110 MB (which turned out to be not enough space to complete my 'yum upgrade'). [Note that some space in /var/yum/cache is *already* being taken up by local copies of repository indexes ("...-primary.sqlite").] .
Comment 1 Mikus Grinbergs 2009-08-26 00:03:46 UTC
A good way to max out the "overhead" storage requirement of yum is to enter 'yum makecache'. I just did that on a Aug 2008 SoaS F11-on-XO1 build, and the resulting output from 'du /var/cache/yum' showed 184 MB was needed.
Comment 2 seth vidal 2009-08-26 16:11:36 UTC
can you include the error message you actually received here?
Comment 3 Mikus Grinbergs 2009-08-26 21:38:43 UTC
(In reply to comment #2) > can you include the error message you actually received here? No error message. [Truly - none whatsoever.] ---- Did testing with virgin (just-installed) F11-on-XO1 build http://dev.laptop.org/~smparrish/xo-1/builds/os6.img . [As part of booting to the command line (in Sugar's 'Terminal' Activity), at home I always enter ' export http_proxy="http://192.168.1.1:8080/" ' and now also ' export https_proxy="https://192.168.1.1:8080/" '.] When I installed it, this build had yum 3.2.23-3.fc1 ; before running the build, I added python-urlgrabber 3.9.0-8.fc12 . For the test, I deliberately mounted /var/cache/yum as a 'tmpfs'. On the XO-1, that created a virtual partition of 110 MB. I then entered 'yum makecache'. The command ran for a while, then simply ended (without any error mesage). The way I could determine that something was wrong was to enter 'df'. That showed that there was no space left in /var/yum cache . Here is a copy of the __only__ output the console ever showed: > 0 [olpc]# yum makecache > fedora/metalink | 15 kB 00:00 > fedora | 3.8 kB 00:00 > fedora/group_gz | 370 kB 00:00 > fedora/filelists_db | 13 MB 00:26 > fedora/prestodelta | 410 B 00:00 > fedora/primary_db | 8.4 MB 00:19 > 0 [olpc]# df > Filesystem 1K-blocks Used Available Use% Mounted on > /dev/mtdblock0 1048576 532096 516480 51% / > tmpfs 110140 0 110140 0% /dev/shm > /tmp 110140 24 110116 1% /tmp > vartmp 110140 8 110132 1% /var/tmp > /dev/mmcblk0p1 7095808 4135648 2599712 62% /media/164f704b-a781-4795-9e2e-0fd812c5f9f5 > /dev/sda5 7874528 4035128 3439384 54% /media/43e6b669-c30b-40d0-8f5c-5901f5f9914a > tmpfs 110140 110140 0 100% /var/cache/yum > 0 [olpc]# You will notice that 'yum' terminates with return code 0. From previous experience, I expect it blew while trying to unpack the filelist.sqlite data for the fedora repo. [There are numerous other repositories listed in /etc/yum.repos.d/ -- what I think happened was that this instance of the yum command never got as far as trying to fetch from those.] .
Comment 4 seth vidal 2009-08-27 17:15:09 UTC
wow,confirmed. That's very very special. looking for what the heck swallows the error.
Comment 5 seth vidal 2009-08-27 20:49:47 UTC
okay. Found it - I love obscure bugs - but I'm sure this has hit a number of people and silently failing is such fun. Fix will be upstream shortly and in before F12-final. Thanks
Comment 6 seth vidal 2009-09-03 16:00:23 UTC
*** Bug 513361 has been marked as a duplicate of this bug. ***
Comment 7 Fedora Update System 2009-09-30 01:38:03 UTC
yum-3.2.24-2.fc11 has been pushed to the Fedora 11 stable repository. If problems still persist, please make note of it in this bug report.
Comment 8 Fedora Update System 2009-10-19 16:44:58 UTC
yum-3.2.24-2.fc10 has been submitted as an update for Fedora 10. http://admin.fedoraproject.org/updates/yum-3.2.24-2.fc10
Comment 9 Fedora Update System 2009-11-04 12:05:14 UTC
yum-3.2.24-2.fc10 has been pushed to the Fedora 10 stable repository. If problems still persist, please make note of it in this bug report.