Hongux sent an email in December to have https://github.com/Wind-River/meta-lat added to the layer index. Tim Orling and others have tried to update the layer index and are seeing problems as described in the irc logs below. [11:46] <vmeson> Several weeks ago, Hongxu asked to have meta-lat added to the layer index but there hasn't been any response. Who is taking care of that now? I'd ask Paul E but he's likely sleeping now. [11:48] <vmeson> We _might_ be able to volunteer someone since we make use of the LI code for our product. [11:49] <RP> vmeson: Paul E is the best we have right now, or maybe moto-timo [11:49] <RP> sakoman: not sure [11:49] <RP> sakoman: probably still early for maint [11:50] <moto-timo> I saw the layer request but forgot to follow up [11:50] <moto-timo> ross is also an admin [11:50] <rburton> hm i thought i pressed approve on that [11:52] <rburton> so the layer index says no unpublished layers to review [11:52] <rburton> but if you go to it, the layer index crashes [11:55] <moto-timo> halstead: I'm getting 'Internal Error' logging in to layer index. can we restart it? [11:56] <abelloni> RP: in https://autobuilder.yoctoproject.org/typhoon/#/builders/87/builds/2862/steps/14/logs/stdio [11:56] <abelloni> you have seen the gdb failure [11:56] <abelloni> but what about Inconsistency detected by ld.so: dl-open.c: 560: dl_open_worker: Assertion `_dl_debug_initialize (0, args->nsid)->r_state == RT_CONSISTENT' failed! [11:56] <abelloni> it seems to be a different issue [11:56] <moto-timo> rburton: yeah, now that I tried I remember the error and decided to wait until after holidays... which means I forgot about it [12:02] <RP> abelloni: I didn't spot that, no. That sounds bad :( [12:02] <RP> abelloni: not sure on a cause :/ [12:03] <rburton> vmeson: i suggest filing a bug that http://layers.openembedded.org/layerindex/branch/master/layer/meta-lat/ produces an error, and copy me/paul/tim [12:03] <rburton> and halstead [12:04] <rburton> and hopefully someone can get to the django logs, as i don't know how to [12:06] <halstead> Yes, I'll restart it and pull logs. [12:06] <halstead> And check on what's up with the docs [12:08] <RP> abelloni: I think that is some kind of comment against all of them now. Hopefully the others I did are more accurate that the assertion one! [12:16] <abelloni> well I checked before triaging ;) [12:20] <RP> abelloni: cool :) [12:55] <vmeson> rburton: I'll do that after I get lunch. [12:57] <rburton> cool [12:57] <rburton> I'll be gone shortly so will look monday unless some beats me [13:03] <halstead> moto-timo: I have captured logs and logged in successfully myself. Watching logs now. I'm curious if you can reproduce now. [13:06] <moto-timo> halstead: https://layers.openembedded.org/layerindex/review/meta-lat/ [13:07] <moto-timo> halstead: as I recall if a layer was already added it used to go to a screen saying so... [13:08] <moto-timo> halstead: sometimes the Internal Server warnings came from Django (which is likely the case here and a bug) or from nginx (which is usually "Bad Gateway") [13:09] <halstead> moto-timo: That page response if big enough it starts buffering to a tmpfile and then I'm not sure why it fails after that. [13:10] <halstead> moto-timo: So you've logged in successfully at least? [13:12] <moto-timo> halstead: indeed [13:12] <moto-timo> halstead: the original reason for today's query was vmeson asking about meta-lat [13:13] <halstead> moto-timo: So disabling the tmpfile it's immediately returning a 500. I'll need to raise the debug log level. [13:13] <moto-timo> halstead: I still don't see it in the layers, so I wonder if the data input was malformed somehow [13:14] <moto-timo> halstead: strange, it's in Layers in the admin interface... [13:14] <halstead> The layer update process is looping with "FileExistsError: [Errno 17] File exists: 'python2' -> '/tmp/tmpc2hwzx78/tmp-glibc/hosttools/python'" [13:14] <moto-timo> oh gee isn't that fantastic? [13:14] <moto-timo> python2 die die die [13:18] <moto-timo> halstead: any thing in the log to indicate which layer is causing that or more hints? [13:19] <halstead> Still reading. [13:19] <moto-timo> this is likely in a part of the code I haven't touched much [15:12] <moto-timo> fwiw I have acces to login to the vm running layer index now. Some of the inner workings are still only in Paul’s head though [15:14] <moto-timo> There are some wonky things from older releases. The ‘python2’ file already exists is come from base.bbclass [15:15] <moto-timo> https://git.yoctoproject.org/poky/tree/meta/classes/base.bbclass?h=zeus#n141
The layer exists in the database (via the 'admin' interface of Django), but something is causing the update layers job to fail.
One problem here is that the layerindex uses "master" to both represent the oe-core branch _and_ the branch of the layer. Since the branch for meta-lat is "main" this breaks the current data model.
Attempt to force the layer to load: $ docker exec <container_id> sh -c "/opt/layerindex/layerindex/update.py -l meta-lat -b main -r" https://layers.openembedded.org/layerindex/updates/33936/
Assigning to Randy to have someone within WR work on it.
Robert, Please rebase our local patches that make sense to be upstreamed and work with Tim to try to get this issue resolved.
Through the admin interface, you can set the actual branch name (if it is different than "master"). Changed layer branch meta-lat: master layer collection to 'lat-layer', actual branch to 'main' Changed layer meta-lat status to 'New'
https://layers.openembedded.org/layerindex/branch/master/layer/meta-lat/
Thanks Tim!
NOTE: we are still working through some issues with layer updates, so this might take a couple days before it finally populates (but that is a meta-data problem).