1 |
-----BEGIN PGP SIGNED MESSAGE----- |
2 |
Hash: SHA1 |
3 |
|
4 |
James Cloos wrote: |
5 |
> Thanks for the replies; my apologies for the delay in repying |
6 |
> back.... |
7 |
> |
8 |
> Based on the replies, I was able to track the problem down. The cdb |
9 |
> "patch" is incompatable with the latest version of portage. Removing |
10 |
> that from /etc/portage/modules allows portage to work. |
11 |
> |
12 |
> That said, there is still the bug of using the filesystem as a |
13 |
> database. It now takes about thirty minutes for any emerge run |
14 |
> to finish reading in /var/cache/edb/dep and most of /var/db/pkg. |
15 |
> |
16 |
> It is imperative that portage use a single file for those databases. |
17 |
> (One per db is fine, in case the above is ambiguous.) |
18 |
|
19 |
In >=portage-2.1.2_rc4-r2 t does that now for installed package (see |
20 |
bug #158931). For /var/cache/edb/dep the sqlite module is available |
21 |
(requires pysqlite or python-2.5 with sqlite support enabled). |
22 |
|
23 |
> This is the same problem that git ran into, leading to pack files, |
24 |
> or that cnews and inn hit, leading to the rotating spool files. |
25 |
> |
26 |
> It may work reasonably well on a headless server, with only a few |
27 |
> packages installed. But on a general purpose workstation with |
28 |
> many packages installed it falls over hard. |
29 |
> |
30 |
> Incidently, the above 30 minute minimum run time is even when |
31 |
> running with --nodeps. |
32 |
|
33 |
That's been fixed for bug #158649. If you make extensive use of |
34 |
PORTDIR_OVERLAY then you will probably want to run emerge --regen |
35 |
after each sync to avoid having to generate cache at other times. |
36 |
|
37 |
> I'll also be posting a bug report about unmerging. Portage is |
38 |
> stat(2)ing *every* file under a directory that might need removal |
39 |
> rather than just checking whether there are any entries in the dir. |
40 |
> Running stat(2) on everything in /usr/lib, /usr/bin, /usr/share/doc |
41 |
> and similar completely thrashes the dirent cache. It is even worse |
42 |
> than running ldconfig when no changes were made in dirs that ldconfig |
43 |
> monitors. |
44 |
> |
45 |
|
46 |
Thanks. Thats fixed in svn r5443. |
47 |
|
48 |
> Thanks for the work; I hope to see performance improvements this year. |
49 |
> |
50 |
> -JimC |
51 |
|
52 |
Zac |
53 |
-----BEGIN PGP SIGNATURE----- |
54 |
Version: GnuPG v1.4.6 (GNU/Linux) |
55 |
|
56 |
iD8DBQFFmeX6/ejvha5XGaMRAugNAJ9fD272i73k7I+OPdLoKwT4L5LutQCbBdbB |
57 |
zV1QY1Wd2icP8AzagGcyg5E= |
58 |
=XLFd |
59 |
-----END PGP SIGNATURE----- |
60 |
-- |
61 |
gentoo-portage-dev@g.o mailing list |