**** BEGIN LOGGING AT Mon Apr 18 02:59:58 2016 Apr 18 06:56:03 good morning Apr 18 08:23:23 morning Apr 18 08:23:44 my recent yocto builds hang with 0 0 CentOS-70-64-mini:36171 yocto-www.yoctopr:https ESTABLISHED Apr 18 08:24:05 any idea what could be wrong there, has anyone seen that? Apr 18 08:24:10 bitbake hangs indefinitely Apr 18 08:37:41 interestingly enough it hangs on package-index Apr 18 08:38:01 why would it contact yocto-www.yoctoproject.org when doing package index?? Apr 18 08:47:38 can it be hanging at PREMIRRORS? Apr 18 08:47:49 I wonder why is it not timing out at all= Apr 18 08:52:39 is bugzilla broken? Apr 18 08:52:41 Software error: Apr 18 08:52:41 DBD::mysql::st execute failed: Got error 28 from storage engine [for Statement "SELECT bugs.bug_id AS bug_id, bugs.bug_severity AS bug_severity, bugs.priority AS priority, bugs.bug_status AS bug_status, bugs.resolution AS resolution, map_product.name AS product, map_component.name AS component, CASE WHEN INSTR(CAST(map_assigned_to.login_name AS BINARY), CAST('@' AS BINARY)) != 0 THEN SUBSTR(map_assigned_to.login_name, 1, INSTR(CAST(ma Apr 18 08:52:42 FROM bugs **** BEGIN LOGGING AT Mon Apr 18 11:00:55 2016 Apr 18 13:34:01 trying to understand why a package split goes wrong. I add ${PN}-bleh to PACKAGES, install a file into ${sysconfdir}/default/bleh and put FILES_${PN}-bleh = "${sysconfdir}/default/bleh". The bleh-file still ends up in the main ${PN} package and my ${PN}-bleh package is just empty. Why does my bleh-file end up in ${PN} when it's in FILES of ${PN}-bleh? Apr 18 13:38:09 because PN takes before PN-bleh Apr 18 13:38:27 (i'm guessing but i expect you use PACKAGES +=, so that's append Apr 18 13:38:39 one easy fix is to use prepend instead Apr 18 13:38:43 (=+) Apr 18 13:39:03 rburton: ah, okay. thanks Apr 18 13:40:07 great, solved it. Apr 18 18:14:53 anyone else seeing this (gnome-terminal:20552): GLib-GIO-ERROR **: No GSettings schemas are installed on the system Apr 18 18:14:53 when invoking devshell? Apr 18 18:15:18 it seems new to me Apr 18 18:51:31 Today I received the password for the wiki account I requested. The thing is, I requested it such a long time ago I don't remember my username anymore. :-) Apr 18 19:48:25 mhm, i have a bbclass that is generating an msdos partition image via a shell function; I am using variables that are not known beforehand (they depend on the number of partitions); in shell i dynamically construct the vars; but somehow they are not passed over from machine.conf? Apr 18 19:48:50 i.e. my machine conf sets DSSIMG_PARTITION_1_SIZE="8192" Apr 18 19:49:12 and in the shell function I would do Apr 18 19:49:13 partvar_image="DSSIMG_PARTITION_${i}_IMAGE" Apr 18 19:49:20 part_image="${!partvar_image}" Apr 18 19:49:31 bitbake -e sees DSSIMG_PARTITION_1_SIZE Apr 18 19:49:52 but echo $DSSIMG_PARTITION_1_SIZE in the bbclass shell function will print nothing Apr 18 19:49:54 why? Apr 18 19:53:55 it's not exported Apr 18 19:54:03 only vars which are exported end up in the task process environment Apr 18 19:54:12 add 'export DSSIMG_PARTITION_1_SIZE' Apr 18 19:54:16 to the metadata Apr 18 19:54:42 also, ${!partvar_image} isn't a portable construct Apr 18 19:54:45 so in machine.conf, instead of DSSIMG_PARTITION_x_SIZE always use export Apr 18 19:54:47 that's a bashism Apr 18 19:54:50 ? Apr 18 19:54:54 ? Apr 18 19:54:56 that sentence didn't make sense Apr 18 19:55:00 :) Apr 18 19:55:07 the vasriable name is irrelevent Apr 18 19:55:11 any variable can be marked as export Apr 18 19:55:19 either export FOO="bar" or FOO="bar"; export FOO Apr 18 19:55:47 but again, ${!partvar_image} isn't appropriate, you'll need to switch that to something portable, i.e. use of eval Apr 18 19:55:51 ok the thing that I probably wanted to ask was where exactly to do the export Apr 18 19:56:00 kergoth: I can rewrite it to eval, that would not be a problem Apr 18 19:56:03 in the metadata, as i said Apr 18 19:56:08 most likely next to hte definition Apr 18 19:56:19 or in the recipe using it, depending Apr 18 19:56:36 in this case i'm assuming the var doesnt' need to be exported everywhere, so i'd do it hwer eyou're using it Apr 18 19:56:46 I had trouble interpreting the term "metadata" Apr 18 19:56:52 i'm not sure how that's confusing Apr 18 19:57:01 bitbake metadata. config files, classes, and recipes Apr 18 19:57:06 ah ok Apr 18 19:57:07 exactly where obviously depends Apr 18 19:57:09 sorry Apr 18 19:59:16 one more question.. recently my builds started to hang with an estabished connection to yocto-www.yoctopr:https ESTABLISHED Apr 18 19:59:26 at least thats what netstat -t says Apr 18 19:59:34 for instance bitbake might hang on package-index Apr 18 19:59:38 and it hangs forever Apr 18 19:59:53 any idea what could this be? Apr 18 20:01:11 no idea, sorry. maybe one of the yocto admin-type-folk might have an idea? i.e. michael halstead? Apr 18 20:01:57 the only greppable reference i could find is a default PREMIRROR setting to yoctoproject.org Apr 18 20:02:13 but then I wonder if bitbake has no timeout there? Apr 18 20:02:19 Jin^eLD, That matches the reverse lookup for the default git server. Is it cloning anything? Apr 18 20:02:41 halstead: I don't think so, ps auxww showed me a bitbake that attempted to do package-index Apr 18 20:02:49 so there was no build activity Apr 18 20:02:57 package-index should be local only Apr 18 20:03:30 it started about last week Apr 18 20:03:39 or mid last week or so Apr 18 20:04:17 We don't mirrors files from that server so I doubt PREMIRROR is involved. Apr 18 20:04:35 I have a nightly build that starts at 3:30 and recently it kept hanging Apr 18 20:04:50 but i do not set the yoct project url anywhere Apr 18 20:05:05 it must be a default setting? I am not even sure what bitbake is trying to do there Apr 18 20:05:40 and by hanging I mean - until it is killed with -9, so really no timeout on this Apr 18 20:07:45 Jin^eLD, I'm not sure what else it would be. Let me know if you have any ideas and I can weigh in. Apr 18 20:08:37 I added -DDD to my build script, maybe I will see more Apr 18 20:08:55 otherwise if you know how or rather where to debug bitbake itself I could try to look into it Apr 18 20:09:15 btw I also had the following when trying to search bugzilla today https://pastebin.mozilla.org/8868158 Apr 18 20:09:29 thought i'd check if someone had a similar issue with bitbake Apr 18 20:17:45 Jin^eLD, I don't see a database error in my logs. Were you able to reproduce the bugzilla issue? Apr 18 20:17:58 I tried it only once, let me try again Apr 18 20:18:04 it was today morning btw Apr 18 20:18:22 I guess between 10 and 11 CEST or so Apr 18 20:18:59 halstead: right now it works without error Apr 18 20:19:21 kergoth: thank you, export did help Apr 18 20:22:25 halstead: btw final message said "For help, please send mail to the webmaster ([no address given]), giving this error message and the time and date of the error." Apr 18 20:22:37 not sure if [no address given] is intentional or a misconfig Apr 18 20:22:41 jus fyi Apr 18 20:22:45 *just Apr 18 20:24:04 Jin^eLD, That's the error message getting generated by a machine behind the reverse proxy that doesn't have a server admin set. (I think.) I can fix that. Apr 18 20:25:23 JaMa, best wishes! Apr 18 20:29:22 Crofton|work: ty! Apr 18 22:02:57 morning all Apr 18 22:02:59 Crofton|work: hey - did you see my question on https://bugzilla.yoctoproject.org/show_bug.cgi?id=9469 Apr 18 22:03:01 ? Apr 18 22:03:11 checking Apr 18 22:03:16 I saw a bunch of emails Apr 18 22:03:50 ah Apr 18 22:03:58 hmm, maybe I postsed here Apr 18 22:04:06 I suspect that sdk will contain pythong Apr 18 22:04:40 ther is mako and cheeath needed in the sdk Apr 18 22:05:58 bluelightning, Apr 18 22:06:28 Crofton|work: with the eSDK you wouldn't add them using TOOLCHAIN_* though Apr 18 22:06:42 no< I do not use that variable Apr 18 22:06:44 hang on Apr 18 22:07:18 wait Apr 18 22:07:19 I do Apr 18 22:07:28 https://github.com/balister/meta-sdr/blob/master/recipes-images/images/native-sdk.inc Apr 18 22:07:41 these are things we need in the sdk Apr 18 22:08:01 sorry for being confusing Apr 18 22:11:24 so, the eSDK contains OE sysroots which should contain everything needed by the SDK; if not, you'd need to use another mechanism to bring those in (for now, do_sdk_depends[depends] += "xyz:do_populate_sysroot abc:do_populate_sysroot") Apr 18 22:12:58 you'll need to adjust how you're adding those items so that they don't get added for the eSDK though Apr 18 22:13:00 you are confusing me, am I doing soething wrong? Apr 18 22:13:13 if you want the eSDK to install properly, yes I'm afraid you are Apr 18 22:13:19 or is this different for the esdk? Apr 18 22:13:34 the eSDK is very differently constructed than the standard SDK Apr 18 22:13:49 is there a way to tell the esdk to ignore this Apr 18 22:14:01 I can't break regualr sdks Apr 18 22:14:05 sure Apr 18 22:15:45 something like this should work: http://pastebin.com/9Tu3BHxL Apr 18 22:16:05 really we ought to iron this out properly Apr 18 22:16:10 I will file a bug Apr 18 22:16:56 weird, why would that be diffeent? Apr 18 22:20:20 because they aren't structured in the same way Apr 18 22:20:50 the eSDK uses OE sysroots, just like OE proper Apr 18 22:21:04 The way you did the variable Apr 18 22:21:09 it doesn't expect all native and target items to be in the SDK sysroots Apr 18 22:21:17 oh right Apr 18 22:21:46 the final task override ensures that the variable is "" when building the eSDK Apr 18 22:21:57 ah Apr 18 22:22:04 I hadn't scolled to the bottom :) **** ENDING LOGGING AT Tue Apr 19 02:59:59 2016