| 1 |
Lately, I've been running into an awful lot of manifest issues in the |
| 2 |
portage tree. Not just on x86, but sparc and mips too. For those that |
| 3 |
may doubt me, add "manifest strict" to FEATURES and see for yourself. |
| 4 |
|
| 5 |
Where these issues are coming from is anyone's guess. Probable suspects |
| 6 |
are those devs not using repoman to check ebuilds into the tree, or we |
| 7 |
might have some kind of corruption running around on the rsync servers, |
| 8 |
or there's some odd bug in portage. |
| 9 |
|
| 10 |
To quote actual evidence, sys-devel/gettext, app-arch/zip, |
| 11 |
app-admin/mirrorselect, and app-admin/gentoolkit gave manifest errors. |
| 12 |
|
| 13 |
on x86, gettext-0.12.1 gives this: |
| 14 |
Calculating dependencies ...done! |
| 15 |
>>> emerge (1 of 1) sys-devel/gettext-0.12.1 to / |
| 16 |
>>> md5 files ;-) gettext-0.12.ebuild |
| 17 |
>>> md5 files ;-) gettext-0.11.3-r1.ebuild |
| 18 |
>>> md5 files ;-) gettext-0.11.1.ebuild |
| 19 |
|
| 20 |
!!! File is corrupt or incomplete. (Digests do not match) |
| 21 |
>>> our recorded digest: 3ae8937c9dd834eff566b1291cd0438a |
| 22 |
>>> your file's digest: 172a26a4a6ab693018ad4b352c7daf36 |
| 23 |
!!! File does not exist: |
| 24 |
/usr/portage/sys-devel/gettext//gettext-0.12.1.ebuild |
| 25 |
|
| 26 |
on sparc: |
| 27 |
>>> emerge (1 of 1) sys-devel/gettext-0.12.1 to / |
| 28 |
>>> md5 files ;-) gettext-0.12.ebuild |
| 29 |
|
| 30 |
!!! File is corrupt or incomplete. (Digests do not match) |
| 31 |
>>> our recorded digest: f1002900097f883163423147dce6d9d9 |
| 32 |
>>> your file's digest: b696d10b2ea72a2dc08b55020f8f79d7 |
| 33 |
!!! File does not exist: /usr/portage/sys-devel/gettext//ChangeLog |
| 34 |
|
| 35 |
(mips gettext is not included as I already re-digested on that machine, |
| 36 |
and an rsync takes ~20mins) |
| 37 |
|
| 38 |
Why these two differ on different machines, I'm not exactly sure on. |
| 39 |
Below are some other snippets between sparc, x86, and mips. |
| 40 |
|
| 41 |
|
| 42 |
|
| 43 |
x86/sparc/mips app-admin/mirrorselect: |
| 44 |
Calculating dependencies ...done! |
| 45 |
>>> emerge (1 of 1) app-admin/mirrorselect-0.6 to / |
| 46 |
>>> md5 files ;-) mirrorselect-0.6.ebuild |
| 47 |
>>> md5 files ;-) mirrorselect-0.3.ebuild |
| 48 |
|
| 49 |
!!! File is corrupt or incomplete. (Digests do not match) |
| 50 |
>>> our recorded digest: 6d748dbe957886a5ca691637d94ec1b3 |
| 51 |
>>> your file's digest: 100d61a5fe0f28e65afbcad758984d6a |
| 52 |
!!! File does not exist: /usr/portage/app-admin/mirrorselect//ChangeLog |
| 53 |
|
| 54 |
In this case, the error is the exact same on all three machines. |
| 55 |
|
| 56 |
|
| 57 |
x86 app-arch/zip: |
| 58 |
>>> emerge (1 of 1) app-arch/zip-2.3-r2 to / |
| 59 |
>>> md5 files ;-) zip-2.3-r1.ebuild |
| 60 |
|
| 61 |
!!! File is corrupt or incomplete. (Digests do not match) |
| 62 |
>>> our recorded digest: 3276fb2669785d06bfb6c2cec123521d |
| 63 |
>>> your file's digest: f56548176b22628e4a08abadbcbad329 |
| 64 |
!!! File does not exist: /usr/portage/app-arch/zip//zip-2.3-r2.ebuild |
| 65 |
|
| 66 |
|
| 67 |
on sparc: |
| 68 |
Calculating dependencies ...done! |
| 69 |
>>> emerge (1 of 1) app-arch/zip-2.3-r2 to / |
| 70 |
>>> md5 files ;-) zip-2.3-r1.ebuild |
| 71 |
|
| 72 |
!!! File is corrupt or incomplete. (Digests do not match) |
| 73 |
>>> our recorded digest: 97e7d71520a7e2309f49e5d7f4951a9b |
| 74 |
>>> your file's digest: 9d33eb6208ea6babfe81ca1113d7258a |
| 75 |
!!! File does not exist: /usr/portage/app-arch/zip//ChangeLog |
| 76 |
|
| 77 |
(masked on mips) |
| 78 |
|
| 79 |
In these cases, two different machines report different md5sums and |
| 80 |
apparent errors in the manifest. |
| 81 |
|
| 82 |
|
| 83 |
I suspect something similar will happen if one tried the same on |
| 84 |
gentoolkit, but this mail is long enough already, so I'll exclude it. |
| 85 |
|
| 86 |
I've been running into these issues quite often, and I suspect several |
| 87 |
reasons. In some cases, it seems like some people aren't using repoman, |
| 88 |
as I'll redigest and commit via repoman, and half an hour later, the |
| 89 |
manifest errors are gone after an rsync. In other cases, like where two |
| 90 |
different machines report different errors in the manifests, I'm |
| 91 |
thinking either corruption on the rsync servers, or a bug in portage |
| 92 |
somewheres. |
| 93 |
|
| 94 |
If someone wants to give this a shot, again, add "manifest strict" to |
| 95 |
FEATURES, and start emerging updates over the next few days. You're |
| 96 |
bound to run into the same issues I'm seeing on multiple machines. |
| 97 |
|
| 98 |
Ideas, thoughts? Or am I just going insane? |
| 99 |
|
| 100 |
--Kumba |
| 101 |
|
| 102 |
|
| 103 |
-- |
| 104 |
"Such is oft the course of deeds that move the wheels of the world: |
| 105 |
small hands do them because they must, while the eyes of the great are |
| 106 |
elsewhere." --Elrond |
| 107 |
|
| 108 |
|
| 109 |
-- |
| 110 |
gentoo-dev@g.o mailing list |
Archives