[illumos-gate] 14972 some diskinfo error messages run into the next line -- Joshua M. Clulow <josh⊙so>

gah.. if only more system daemons linked against libumem by default

they could!

yes.. then i would have saved myself more time than I care to admit on my own bug :)

andyf: hi, did you see similiar issues with pkg server on bloody? pastebin.com/S7ptyFiC

looks like it's not happy with cherrypy (whatever that is), but i can't find related package (?)

That rings a faint bell.. I would try running this small test script to check that cherrypy is loading properly (and it might throw out a message about the real problem)

Knowing cherrypy, they could have added yet another jaraco.* dependency...

ModuleNotFoundError: No module named 'cherrypy'

looking through history, it now ships with pkg itself, so i need to tell python somehow about it?

Ah yes, sorry, try paste.ec/paste/4Uaz2s8u#kcUCbucCwAbPwunzq8+J8qQmyKXF-MNCVO+j/inoptk

now it's not happy with pkg_resources: ModuleNotFoundError: No module named 'pkg_resources'

this is likely the real error here? i see that cherrypy returns "unknown" as version

(which breaks pkg.depotd check)

yes, that's probably the real issue. I wonder where that comes from (pkg_resources) - it sounds built in

part of setuptools-310 - is that installed?

Could be a missing dependency in pkg here

installing (it wasn't)

w/39

yay, that fixed it, thanks!

Great, I'll get it added as a dependency for pkg, and find out why it was not automatically detected

[illumos-gate] 14992 dead code in desctbls.c -- Dan Cross <cross⊙oc>

tsoome, andyf: Unfortunately I had to make code.illumos.org/c/illumos-gate/+/2355 more complex, when you get a moment to have a look

→ CODE REVIEW 2355: 14978 ZFS autoexpand property should work for root pools (NEW) | illumos.org/issues/14978

jclulow aye, I already did read it, just wanted to get back to it a bit later:)

Thanks!

Basically the point is, we prepare zfe_mod to have valid device names beforehand?

jclulow - I'll take a look

External Bug is nice;) /me likes

jclulow - so the pool import kicks off the autoexpand task, and that sends a DLE sysevent for all of the VDEVs, and now the sysevent handler does a refresh before re-opening the pool and calling vdev_online().. did I get that about right?

(and TIL that pools with autoexpand get opened twice during import..)

Yes the DLE event is always handled by the module in syseventd today

And now we're doing the refresh that "zpool status" does

Thanks, makes sense

right. of course you get white space noise while preparing the RTI mail.