Bug 1047087 - make the creation of the .solv files more atomic
Summary: make the creation of the .solv files more atomic
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Fedora
Classification: Fedora
Component: hawkey
Version: 20
Hardware: x86_64
OS: Unspecified
low
medium
Target Milestone: ---
Assignee: Ales Kozumplik
QA Contact: Fedora Extras Quality Assurance
URL: https://retrace.fedoraproject.org/faf...
Whiteboard: abrt_hash:8c40aac529dcf08d1e445235f1e...
: 1051812 1066783 (view as bug list)
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2013-12-28 23:05 UTC by José Antonio
Modified: 2016-05-30 13:10 UTC (History)
9 users (show)

Fixed In Version: hawkey-0.4.11-1.fc20
Clone Of:
Environment:
Last Closed: 2014-02-25 07:49:33 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
File: backtrace (43.22 KB, text/plain)
2013-12-28 23:05 UTC, José Antonio
no flags Details
File: cgroup (164 bytes, text/plain)
2013-12-28 23:05 UTC, José Antonio
no flags Details
File: core_backtrace (10.16 KB, text/plain)
2013-12-28 23:05 UTC, José Antonio
no flags Details
File: dso_list (9.35 KB, text/plain)
2013-12-28 23:06 UTC, José Antonio
no flags Details
File: environ (72 bytes, text/plain)
2013-12-28 23:06 UTC, José Antonio
no flags Details
File: exploitable (82 bytes, text/plain)
2013-12-28 23:06 UTC, José Antonio
no flags Details
File: limits (1.29 KB, text/plain)
2013-12-28 23:06 UTC, José Antonio
no flags Details
File: maps (48.84 KB, text/plain)
2013-12-28 23:06 UTC, José Antonio
no flags Details
File: open_fds (589 bytes, text/plain)
2013-12-28 23:06 UTC, José Antonio
no flags Details
File: proc_pid_status (889 bytes, text/plain)
2013-12-28 23:06 UTC, José Antonio
no flags Details
File: var_log_messages (340 bytes, text/plain)
2013-12-28 23:06 UTC, José Antonio
no flags Details

Description José Antonio 2013-12-28 23:05:49 UTC
Version-Release number of selected component:
dnf-0.4.9-1.fc20

Additional info:
reporter:       libreport-2.1.10
backtrace_rating: 4
cmdline:        /usr/bin/python /usr/bin/dnf -v makecache timer
crash_function: solvable_copy_cb
executable:     /usr/bin/python2.7
kernel:         3.12.5-302.fc20.x86_64
runlevel:       N 5
type:           CCpp
uid:            0

Truncated backtrace:
Thread no. 1 (10 frames)
 #0 solvable_copy_cb at /lib64/libsolvext.so.0
 #1 repodata_search at /lib64/libsolv.so.0
 #2 repo_add_rpmdb at /lib64/libsolvext.so.0
 #3 repo_add_rpmdb_reffp at /lib64/libsolvext.so.0
 #4 hy_sack_load_system_repo at /lib64/libhawkey.so.0
 #5 load_system_repo at /usr/lib64/python2.7/site-packages/hawkey/_hawkeymodule.so
 #6 PyEval_EvalFrameEx at /lib64/libpython2.7.so.1.0
 #7 PyEval_EvalCodeEx at /lib64/libpython2.7.so.1.0
 #8 PyEval_EvalFrameEx at /lib64/libpython2.7.so.1.0
 #9 PyEval_EvalCodeEx at /lib64/libpython2.7.so.1.0

Comment 1 José Antonio 2013-12-28 23:05:54 UTC
Created attachment 842806 [details]
File: backtrace

Comment 2 José Antonio 2013-12-28 23:05:55 UTC
Created attachment 842807 [details]
File: cgroup

Comment 3 José Antonio 2013-12-28 23:05:58 UTC
Created attachment 842808 [details]
File: core_backtrace

Comment 4 José Antonio 2013-12-28 23:06:00 UTC
Created attachment 842809 [details]
File: dso_list

Comment 5 José Antonio 2013-12-28 23:06:01 UTC
Created attachment 842810 [details]
File: environ

Comment 6 José Antonio 2013-12-28 23:06:03 UTC
Created attachment 842811 [details]
File: exploitable

Comment 7 José Antonio 2013-12-28 23:06:05 UTC
Created attachment 842812 [details]
File: limits

Comment 8 José Antonio 2013-12-28 23:06:07 UTC
Created attachment 842813 [details]
File: maps

Comment 9 José Antonio 2013-12-28 23:06:09 UTC
Created attachment 842814 [details]
File: open_fds

Comment 10 José Antonio 2013-12-28 23:06:12 UTC
Created attachment 842815 [details]
File: proc_pid_status

Comment 11 José Antonio 2013-12-28 23:06:14 UTC
Created attachment 842816 [details]
File: var_log_messages

Comment 12 Ales Kozumplik 2014-01-02 17:46:21 UTC
hmmm, there's no coredump really in here and I've been so far unable to map back the traceback's addresses to source files/lines.

One thing that stands out: the crash occurs in solvable_copy_cb() which is what got affected by mls last patch in the latest build, a8e47f15b81002c721fe164885d809562c5d3e7a, the one that fixes issues with package checksums if updating rpmdb cache with repo_add_rpmdb_reffp(). Michael, is it possible the issue is in this?

Comment 13 Ales Kozumplik 2014-01-02 19:26:44 UTC
here's a better backtrace:

https://retrace.fedoraproject.org/faf/reports/310947/

That would indicate the crash in dirpool_parent() (still indirectly called from solvable_copy_cb)

Comment 14 Ales Kozumplik 2014-01-02 19:35:25 UTC
reporter, we might need the full coredump file. can you please provide us one, by doing:

$ ulimit -c unlimited
$ /usr/bin/python /usr/bin/dnf -v makecache

This will create (in case of crash) core.* file in the local directory, I'd like to ask you to upload it somewhere and paste the link here, if possible. Thank you.

Comment 15 Ales Kozumplik 2014-01-02 20:20:34 UTC
related abrt issue why the backtrace is no good here: bug 1047980

Comment 16 Michael Schröder 2014-01-07 12:30:39 UTC
I don't think it has anything to do with the latest change, as this is about the filelist, not a checksum. My guess is that the old solv file is corrupt, the directory index of one of the files seems to be out of the valid range.

Comment 17 Ales Kozumplik 2014-01-08 12:37:22 UTC
Not sure what to do here, we shouldn't crash on any input .solv file, but it's too rare (possibly because of hw error etc.) to be worth actively chasing and fixing at the moment. Let's just keep it open for a record, perhaps more users will experience this.

Also, reporter, if you still see this: can you upload /var/cache/dnf/x86_64/20/@System.solv to this? Thank you.

Comment 18 Ales Kozumplik 2014-01-13 06:34:15 UTC
similar bug with somewhat a better traceback: bug 1051917.

Comment 19 Ales Kozumplik 2014-01-22 09:17:26 UTC
uploaded the core http://akozumpl.fedorapeople.org/core.9039

Comment 20 Ales Kozumplik 2014-01-22 09:18:40 UTC
scratch comment 19, wrong bug.

Comment 21 Ales Kozumplik 2014-02-06 12:09:12 UTC
Discussing this with Michael we decided the best course of action for now is eliminate the possibility of keeping an unfinished solv file in the cache. To prevent that we will provide functionality that first stores the generated solv file into a temporary file and only when that is successfully finished rename it to the final name.

Comment 22 Ales Kozumplik 2014-02-10 15:39:51 UTC
Fixed upstream by 9290a1c, will be part of the 0.4.10 release.

Comment 23 Ales Kozumplik 2014-02-10 15:40:12 UTC
Fixed upstream by 9290a1c, will be part of the 0.4.10 release.

Comment 24 Fedora Update System 2014-02-17 14:37:14 UTC
hawkey-0.4.10-1.fc20 has been submitted as an update for Fedora 20.
https://admin.fedoraproject.org/updates/hawkey-0.4.10-1.fc20

Comment 25 Fedora Update System 2014-02-18 13:44:36 UTC
Package hawkey-0.4.10-1.fc20:
* should fix your issue,
* was pushed to the Fedora 20 testing repository,
* should be available at your local mirror within two days.
Update it with:
# su -c 'yum update --enablerepo=updates-testing hawkey-0.4.10-1.fc20'
as soon as you are able to.
Please go to the following url:
https://admin.fedoraproject.org/updates/FEDORA-2014-2700/hawkey-0.4.10-1.fc20
then log in and leave karma (feedback).

Comment 26 Ales Kozumplik 2014-02-19 10:23:19 UTC
*** Bug 1066783 has been marked as a duplicate of this bug. ***

Comment 27 Fedora Update System 2014-02-24 17:52:52 UTC
hawkey-0.4.11-1.fc20 has been submitted as an update for Fedora 20.
https://admin.fedoraproject.org/updates/hawkey-0.4.11-1.fc20

Comment 28 Fedora Update System 2014-02-25 07:49:33 UTC
hawkey-0.4.10-1.fc20 has been pushed to the Fedora 20 stable repository.  If problems still persist, please make note of it in this bug report.

Comment 29 Fedora Update System 2014-02-27 04:13:28 UTC
hawkey-0.4.11-1.fc20 has been pushed to the Fedora 20 stable repository.  If problems still persist, please make note of it in this bug report.

Comment 30 Ales Kozumplik 2014-03-10 15:06:06 UTC
*** Bug 1051812 has been marked as a duplicate of this bug. ***

Comment 31 Michael Schröder 2016-05-30 13:10:11 UTC
(clearing needinfo)


Note You need to log in before you can comment on or make changes to this bug.