Summary: | sys-apps/portage-2.2.20.1 and emerge --fetchonly binary packages download loop | ||
---|---|---|---|
Product: | Portage Development | Reporter: | marco |
Component: | Binary packages support | Assignee: | Portage team <dev-portage> |
Status: | UNCONFIRMED --- | ||
Severity: | normal | CC: | zerochaos |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
packages sample file
debug patch Extract of Packages from the Builder Extract of Packages from the server emerge log after patch |
Description
marco
2015-10-21 10:36:31 UTC
The binary packages are compared using a tuple generated by the _instance_key_multi_instance method here: https://gitweb.gentoo.org/proj/portage.git/tree/pym/portage/dbapi/virtual.py?h=v2.2.20.1#n55 Can you attach some of the duplicate entries from you $PKGDIR/Packages file? That will probably show us why _instance_key_multi_instance is returning a different result. If the package mtime changes or is not preserved for some reason, then that could trigger it. Created attachment 415140 [details]
packages sample file
Packages file added (In reply to marco from comment #2) > Created attachment 415140 [details] > packages sample file All of those entries appear to be identical. We should look at the server-side entry for the same package. Is it identical to these entries? Created attachment 415214 [details, diff]
debug patch
This patch will make emerge output some useful debugging information to stderr. Sample usage:
emerge -pvg ncurses > debug.log 2>&1
Created attachment 415222 [details]
Extract of Packages from the Builder
Created attachment 415224 [details]
Extract of Packages from the server
Created attachment 415226 [details]
emerge log after patch
(In reply to Zac Medico from comment #5) > Created attachment 415214 [details, diff] [details, diff] > debug patch > > This patch will make emerge output some useful debugging information to > stderr. Sample usage: > > emerge -pvg ncurses > debug.log 2>&1 I have attached the debug.log file. Thanks for the support (In reply to marco from comment #8) > Created attachment 415226 [details] > emerge log after patch This shows that the instance key contains a None value for the BUILD_ID, while the local instance has BUILD_ID: 1 in the Packages file, so that explains why the comparison is failing. I'll have to investigate some more to find out what causes this. *** Bug 563820 has been marked as a duplicate of this bug. *** Does is make any difference if you remove the cached copy of the remote Packages file? You can remove all cached copies with this command: rm -rf /var/cache/edb/binhost (In reply to Zac Medico from comment #12) > Does is make any difference if you remove the cached copy of the remote > Packages file? You can remove all cached copies with this command: > > rm -rf /var/cache/edb/binhost Previously i tried to remove /usr/portage/packages/ but without success. Perfect , once removed the edb/binhost cache all is ok . Do you need something else to investigate ? (In reply to marco from comment #13) > Perfect , once removed the edb/binhost cache all is ok . Great! > Do you need something else to investigate ? Now that we have isolated the problem to the cache, I'll focus my attention on that area, and then I'll let you know if I need anything else. @zerochaos: Does removing the cache also solve it for you? |