Build failing for i386?

[Couch Potato]

Hi Guys,

Been playing with the builds on Virtual Box for a while now without problem, but for the last week or so "buildworld" has been failing...

Code:

Package is not yet built: mail_p5-Net-SMTP-SSL
Package is not yet built: lang_p5-Error
Package is not yet built: devel_git
Package is not yet built: devel_ctags
Automatically building a * * F A T * * image so we can build ports
00:00:00 # NanoBSD image FreeNAS-8r7765-i386 build starting
00:00:00 ## Clean and create object directory (/usr/local/freenas/trunk/obj.i386)
00:00:01 ## Construct build make.conf (/usr/local/freenas/trunk/obj.i386/make.conf.build)
00:00:01 ## run buildworld
00:00:01 ### log: /usr/local/freenas/trunk/obj.i386/_.bw
build/do_build.sh: ERROR: FreeNAS build FAILED; please check above log for more details

Tailing _.bw I get...

Code:

===> gnu/usr.bin/texinfo/info (installincludes)
===> gnu/usr.bin/texinfo/infokey (installincludes)
===> gnu/usr.bin/texinfo/install-info (installincludes)
===> gnu/usr.bin/texinfo/texindex (installincludes)
===> gnu/usr.bin/texinfo/doc (installincludes)
1 error
*** Error code 2
1 error
*** Error code 2
1 error

This is beyond my skillset (at the moment) anyone have any bright ideas why?

Ta!

Andy B.

 

[joeschmuck]

Andy,
Three things that will cause failures during the builds that I know of are...

1) You didn't delete obj.i386 before running do_build
2) The developers haven't updated one of the files related to the build process due to changes they are making
3) The CSUP server cannot serve up all the needed files

Items 1 & 2 are the most likely cause why a build fails. Item 3 is possible but if you have used that server several times before, it's probably not the failure.

If you don't think any of these are it, toss out another message and give some details on what build number and if yo have enough virtual hard drive space, etc...

EDIT: Have you tried an amd64 build, just to see if it builds?

-Mark

 

[Couch Potato]

Hi Mark,

The current failure is on 8r7765 though I've had similar problems from about 7760 onwards...

I've deleted obj.i386 and tried again to no avail, I even completely deleted trunk and rebuilt from scratch with the same results.

Unfortunately I can't try an amd64 build (I'm on an old 386 Laptop at the moment) and I don't believe I can build a 64 bit image on a 32 bit machine.

I'm hedging my bets that Option 2 (above) is the problem

The Virtual machine I'm running has 1G of ram and 20G of disk

fdisk reports:-

Code:

freebsd# fdisk
******* Working on device /dev/ad0 *******
parameters extracted from in-core disklabel are:
cylinders=41610 heads=16 sectors/track=63 (1008 blks/cyl)

Figures below won't work with BIOS for partitions not in cyl 1
parameters to be used for BIOS calculations are:
cylinders=41610 heads=16 sectors/track=63 (1008 blks/cyl)

Media sector size is 512
Warning: BIOS sector numbering starts with sector 1
Information from DOS bootblock is:
The data for partition 1 is:
sysid 165 (0xa5),(FreeBSD/NetBSD/386BSD)
    start 63, size 41942817 (20479 Meg), flag 80 (active)
        beg: cyl 0/ head 1/ sector 1;
        end: cyl 1023/ head 15/ sector 63
The data for partition 2 is:
<UNUSED>
The data for partition 3 is:
<UNUSED>
The data for partition 4 is:
<UNUSED>

and df -h:-

Code:

freebsd# df -h
Filesystem     Size    Used   Avail Capacity  Mounted on
/dev/ad0s1a    989M    176M    734M    19%    /
devfs          1.0K    1.0K      0B   100%    /dev
/dev/ad0s1e    989M    4.7M    905M     1%    /tmp
/dev/ad0s1f    7.9G    2.1G    5.2G    29%    /usr
/dev/ad0s1d    5.7G     18M    5.3G     0%    /var
freebsd#

I don't think it's a space issue (the df -h is with trunk intact, deleting trunk ups the free space on /usr to 6.4G.

I'd hasten to add that I'm new to both FreeBSD and FreeNAS (ex Solaris admin) so may well be missing something else...

Cheers!

Andy B.

 

[Durkatlon]

I have the exact same problem, and I even rebuild the FreeBSD VM from scratch (for other reasons) and the same error occurred again. That would be the cleanest build starting point I can think of :D. I was able to build in the past but haven't been able to build i386 for a while now. I am starting to suspect the cvsup server, but I haven't looked into it in detail yet (working on building FreeNAS 0.7 i386 right now, also problems there... :))

 

[Couch Potato]

Funnily enough I've also just rebuilt the FreeBSD VM from scratch, upping the disk size from 20G to 40G in case that was the problem - it wasn't - :(

If I get time over the next few days I'll delve back into some older versions and see if I can work out where/when it broke...

Andy B

 

[joeschmuck]

I'm setting up my VM now. It takes a few minutes because I use a 16GB RAM Disk to speed things up a little. I'll post my results when I'm done.

-Mark

 

[joeschmuck]

I just submitted ticket #785 for the problem. Needless to say I was unable and I tried a few ways to create it. I changed the CSUP, how I start the build, and a few litle things I didn't think would help, which of course didn't. I have attached my _.bw log file. Feel free to attach your own log file to the ticket. I renamed it from _.bw to BuildWorld.log and then ZIP the file because it was too big uncompressed. Please call yours something slightly different as to not replace mine.

-Mark

 

[Durkatlon]

Thanks for confirming the build error. As soon as I boot up my FreeBSD8.2 VM I'll grab my log and attach it to your ticket. Currently all the RAM on my VMWare box is taken up by the FreeBSD7.2 VM used to compile FreeNAS 0.7.

OFF TOPIC: I've decided to try and finally get my FreeNAS 0.7 machines to use all 4GB of the RAM they have. I'm compiling build 5127 of FreeNAS 0.7 since that is what my boxes actually run. This was the final release of "Shere" (0.7.1). It's quite challenging to get this old build to compile. So far I've had to edit about 10 Makefiles, grabbed 2 tar.gzs by hand because they were no longer available in the location ports knows about, and made two changes to ffmpeg.cpp because it wouldn't compile against the libraries that are pulled in. :D

 

[Couch Potato]

Hi Mark,

Thanks for confirming the build error, at least it's not me cracking up :)

I've re-run the build and pinned the log to the ticket - the more the merrier!

Cheers!

Andy B

 

[Couch Potato]

Seems to be resolved now... thanks gcooper and joeschmuck

Just an observation, but it seems odd that the problem was shell related - it would seem that the "default" shell for FreeBSD is csh so wouldn't you work in that shell for testing or at least document that the shell needs to be changed?

 

[Durkatlon]

Wait, what? What was the fix?

 

[Couch Potato]

gcooper posted this in ticket #785...

Figured out what the problem was. Annoying that ${SHELL} expands to /bin/csh in the build, which breaks set -e in a script I just added via the patch >_>.

So I'm guessing that its the differences between the interpretation of "set" in the Bourne? and C shells?

I may have got this wrong - I havent touched Unix for many years and used to spend most of my time in the korn shell anyway! :)

Cheers!

Andy B