| View previous topic :: View next topic |
| Author |
Message |
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3687
Location: Rasi, Finland
|
 Posted: Tue Jul 30, 2024 10:30 am Post subject: Oddly long compile (install?) times |
 |
|
| the relevant part: | panther-box ~ # qlop gentoolkit
2024-01-27T20:29:57 >>> app-portage/gentoolkit: 27s
2024-02-01T18:16:47 >>> app-portage/gentoolkit: 38s
2024-03-12T16:04:57 >>> app-portage/gentoolkit: 30s
2024-04-11T13:21:07 >>> app-portage/gentoolkit: 33s
2024-05-16T03:53:08 >>> app-portage/gentoolkit: 4:53:03
2024-07-30T01:08:46 >>> app-portage/gentoolkit: 4:40:49 |
See? This is obviously wrong. There are some other packages that seem to have way too long compile times.
I have suspicion: | *snip*: | 2024-07-30T01:03:54 >>> net-misc/modemmanager-1.20.6: 4′52″
2024-07-30T01:08:46 >>> app-portage/gentoolkit-0.6.5: 4:40:49
2024-07-30T01:08:50 >>> sys-devel/llvm-18.1.8-r1: 4:40:56
2024-07-30T05:49:46 >>> sys-devel/llvm-toolchain-symlinks-18: 15s
2024-07-30T05:50:01 >>> sys-devel/llvmgold-18: 16s
2024-07-30T05:50:17 >>> dev-util/spirv-llvm-translator-18.1.2: 8′13″
2024-07-30T05:58:30 >>> sys-devel/clang-18.1.8: 4:50:41
2024-07-30T10:49:11 >>> sys-devel/clang-toolchain-symlinks-18: 14s |
... in this case gentoolkit was being held in a state between compilation and installation until llvm was ready.
I wonder if this is a bug in portage or just a small parsing bug in qlop..?
This is a "remote" machine on in my home, so I don't have access to in in next few days.
_________________
..: Zucca :..
| My gentoo installs: | init=/sbin/openrc-init
-systemd -logind -elogind seatd |
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
 |
eccerr0r
Watchman
Joined: 01 Jul 2004
Posts: 9822
Location: almost Mile High in the USA
|
| Posted: Tue Jul 30, 2024 2:06 pm Post subject: |
|
|
Yeah I think this is due to running emerge --jobs and due to install ordering it chose to install it after llvm finished for whatever reason... I totally discount any build times in /var/log/emerge.log due to this if I do emerge --jobs ... at the very least, and yes this has happened, I've had emerge --jobs build qtwebengine and firefox at the same time. That's definitely going to hurt.
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 22618
|
| Posted: Tue Jul 30, 2024 2:13 pm Post subject: |
|
|
| In my opinion, this represents a design limitation in creating tools like qlop. Portage logs enough information to get this right in simple cases, but if you really want accurate data, then you need Portage to log the data in a way that does not require qlop to infer anything, and instead allows qlop to be a glorified grep + reformatting tool. |
|
| Back to top |
|
|
eschwartz
Developer
Joined: 29 Oct 2023
Posts: 214
|
| Posted: Tue Jul 30, 2024 3:56 pm Post subject: |
|
|
FEATURES=merge-wait is relevant here as well -- note the description: "it can be disabled in order to trade the possibility of random build failures for greater parallelism."
See also https://bugs.gentoo.org/663324
Of course, the "build qtwebengine and firefox at the same time" case is broadly speaking unsolvable since in such a scenario every package genuinely takes longer to build since they fail to get an exclusive lock on system resources. |
|
| Back to top |
|
|
|
Portage & Programming
