Bug 1211891
| Summary: | If an error occurs dnf don't resume work and re-redownload everything. | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Fedora] Fedora | Reporter: | Mikhail <mikhail.v.gavrilov> | ||||
| Component: | dnf | Assignee: | Packaging Maintenance Team <packaging-team-maint> | ||||
| Status: | CLOSED CURRENTRELEASE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> | ||||
| Severity: | unspecified | Docs Contact: | |||||
| Priority: | unspecified | ||||||
| Version: | 22 | CC: | covex, jmontleo, jones.peter.busi, jonha87, jsilhan, jzeleny, mikhail.v.gavrilov, mluscon, mustafa1024m, packaging-team-maint, pnemade, tim.lauridsen, vmukhame | ||||
| Target Milestone: | --- | Keywords: | Reopened | ||||
| Target Release: | --- | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Whiteboard: | |||||||
| Fixed In Version: | Doc Type: | Bug Fix | |||||
| Doc Text: | Story Points: | --- | |||||
| Clone Of: | Environment: | ||||||
| Last Closed: | 2015-10-19 12:24:39 UTC | Type: | Bug | ||||
| Regression: | --- | Mount Type: | --- | ||||
| Documentation: | --- | CRM: | |||||
| Verified Versions: | Category: | --- | |||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||
| Embargoed: | |||||||
| Attachments: |
|
||||||
|
Description
Mikhail
2015-04-15 07:41:40 UTC
Same issue. Get hung up on a missing gpg key, import, try to install again, redownload everything again. Thanks for the report. DNF tries to clean up after any job. Does it happen with `keepcache=True` option set in /etc/dnf/dnf.conf? Thanks, with option "keepcache=true" DNF work as expected. I think you should enable this option by default. Who needs to clear the cache will always be able to type "dnf clean all" I'm afraid that a lot of users will complain if we wouldn't remove downloaded packages automatically. Actually there is the "local" plugin for those people who have unlimited disk space. We could rather "keepcache" only in case of an error. But it's questionable which concrete error should it be. (In reply to Radek Holy from comment #4) > We could rather "keepcache" only in case of an error. But it's questionable > which concrete error should it be. 1. Error downloading packages 2. missing gpg key Yeah. I should rather say: We should carefully consider which errors should it be. I agree with those two. (Hopefully, librepo allows us to recognize them.) I was just installing 420MB of updates and after the download finished (it took some time because I have rather bad internet) dnf complained that the disk is full and aborted the installation. So I deleted some files and started the update again, and dnf redownloads everything again. This is stupid. Other package managers keep the files in cache somewhere and delete them only if they aren't needed anymore (because the update succeeded or whatever) or after some time has passed (I don't know the details). TLDR, a full disk error should also cause the cache to be kept. *** Bug 1217876 has been marked as a duplicate of this bug. *** I wonder how helpful (and difficult) would it be to remove cache only during the next run... In combination of this with the cache sharing service (which is being developed these days, I believe) it could even result in no need for --keepcache any more. *** This bug has been marked as a duplicate of bug 1218415 *** But this is not the same as that package, I don't believe it is a duplicate. When I downloaded 200 MBs of updates I don't want it to go if I press no, or a problem in gpg checking. As the duplicated bug moved completely out of scope, I do reopen this one, as this behavior of DNF is plain wrong. This bothers me every then and now as dnf deletes all packages downloaded for running transaction on transaction failures. This has nothing to do with keepcache option as it was in the yum - this meant to not remove cache on successful completition. We do want the same behavior as yum did it. Why the hell should I redownload 4 times 1GB of updates that takes 15minuts just because it a) was not able to install kernel in /boot due to not enought space? b) was missing gpg signature for certain package c) mirrors have failed due to dropout in the internet connection d) ... Are you aware that we reworked this behaviour of dnf?
keepcache
boolean
Keep downloaded packages in the cache. If set to False and packages have not been installed they will still persist until next successful transaction. The default is False.
|