« first day (2456 days earlier)      last day (2796 days later) » 

08:54
Since upgrade to stretch, I'm seeing the following email after every reboot. Sometimes multiple times:
> orwell Communications with UPS 1 lost
I assume this is coming from apscupsd. A configuration change, perhaps?
@terdon Hi, it looks like this one was never acted on...
13
A: Why do we have both LVM and an LVM2 tags?

GillesI don't recall seeing a question about LVM1 on this site. It was already obsolete by the time this site started. So for all intents and purposes lvm and lvm2 are currently equivalent. I favor merging them, and making lvm the main tag. (Only moderators can do this.) If an incompatible LVM3 comes...

I think (as @Gilles says) it makes sense to do a merge. Thoughts?
I do think one needs to plan ahead what should be done if arrives.
09:10
@FaheemMitha Done.
@terdon Thank you. What's the plan re ? Assuming it arrives at some point.
@FaheemMitha No idea. What Gilles suggested there seems reasonable though.
@terdon I'm not exactly sure what he suggested...
> If an incompatible LVM3 comes along, we can start using lvm3 for it. Then make lvm2 the main tag in the lvm/lvm2 pair and remove the synonym.
Then again, before is released, a giant asteroid might hit Earth, and we'll all perish like the dinosaurs.
@terdon Yes, I know. See my comment.
Or, before is released, the President of the United States might start World War 3, and we'll all perish like utter idiots.
09:22
@FaheemMitha I don't really understand what you're missing. Obviously, if lvm3 and 2 are not incompatible, we'll make them synonyms. If they are, we'll make an lvm3 tag.
@terdon Would the questions currently tagged become if lvm3 becomes the new version of lvm?
@FaheemMitha If they're compatible, we would still only have one and both and would be synonyms of .
If it is significantly different and requires its new tag, would be separate and and would be synonyms.
@terdon Hmm. Does the current lvm need to be tagged as lvm? And if lvm3 becomes current, it would be the new lvm.
Though maybe not. I don't really know.
09:38
I don't understand what you mean. I have merged and and made the latter a synonym of the former. This means no longer exists and if anyone tries to use it, it will automatically be converted to .
So all questions that were tagged with are now tagged with instead.
Now, if an LVM3 comes along and doesn't need its own tag, we just keep using . If it does need its own tag, we can create .
@terdon Yes. I understand. Though in that case, LVM3 questions will not be tagged , correct?
@FaheemMitha No, not if they're significantly different but why would that be the case?
Let's just cross that bridge if and when we get to it.
09:54
@terdon will this question be on-topic on U&L?
@Pandya Yes, but as they told you there, the answer will almost certainly be no.
@terdon Hmm... ok
@terdon Because LVM3 questions are still LVM. To put it another way, if a user hypothetically tags a LVM3 question , would it be removed?
@terdon <Shrug.> Ok.
Why does Ethernet have 1048 fragments?
@Pandya Broadcom doesn't have free software support, afaik. The usual problem is that the manufacturers don't release the specs.
You could always check with the manufacturers directly. Assuming they reply, that is.
If you get support in India, I'd hang up.
@Pandya If possible, I try to make sure that I purchase hardware that works with free software. It's usually more reliable, and often the hardware itself is better.
Of course, this isn't always possble. E.g. laptops.
10:14
Also, @Pandya if chili555 from AU couldn't help you, I doubt anyone can. He's very knowledgeable about the wireless driver situation.
@FaheemMitha yeah. that's good point.
He's the only one with the gold wireless badge and the undisputed local expert on wireless.
@terdon who's chilli555? I discussed with other user.
answer before question!
@FaheemMitha I'll also purchase FSF certified laptop once.
@Pandya Heh, yes, I'm just that good :P
2
10:23
@Pandya You could, but it's not necessary to go to those lengths, I think. Just research hardware before you buy it. If everyone did that, there would be less Linux hardware questions on SE.
Btw, trying to install proprietary driver for long time ^^
@FaheemMitha yes but I may like to go to that length!
@Pandya Ok. :-)
Trisquel release based on 16.04 got delayed for 1 year.
 
1 hour later…
11:46
Does anyone know why SE question titles have to be at least 15 chars?
@FaheemMitha Presumably to avoid stupid titles like "Help!"
And to push you into writing useful, descriptive titles.
12:01
what's the command-line alternative to install this driver? because GUI is not responding after long time and I can't see what is happening in background.
12:51
Anyone know how populist badge are awarded ? I just got one in SuperUser for a post accepted 15 month ago, and having not receive too much points recently.
13:02
@Pandya What driver?
27
Q: Is the Populist badge broken? Or have I misunderstood?

RichieHindleIn this question: First name, middle name, last name. Why not Full Name? my answer has 34 votes and the accepted answer has 10. But no sign of a Populist badge - should I have one? Edit: The accepted answer on that question has received another upvote, so I now have the badge. But either the d...

But whether this is still accurate or not, who knows.
@terdon It's possible to have useful, descriptive titles that are shorter than 15 chars.
And anyone who uses "Help" as a title deserves a sound ducking.

Wifi problem solved

1 hour ago, 52 minutes total – 48 messages, 5 users, 0 stars

Bookmarked 22 secs ago by Pandya

@Pandya You could just say that you figured it out...
I must say that AU folks seem helpful. In the average forum they mostly just ignore you. Or you get the occasional sideways hostile glance.
Like a saloon/bar in the Wild West of legend.
@FaheemMitha Yes, but harder. I assume they added a minimum character limit to force people to use descriptive titles.
@terdon No doubt. Still occasionally aggravating.
My titles tend towards the wordy, so it isn't usually an issue.
But some days I'm in a short, pithy mood.
Ah, the old short and pithy mood, yes. :P
 
2 hours later…
15:01
I realise this is a vague question, but does LVM know or care what kind of device/structure it's sitting on top of? Or does it just tag it with some metadata and carry on?
 
1 hour later…
16:19
@FaheemMitha what do you mean? It cares that whatever it's on is a block device. Other than that, I think it might do a little to make alignment work.
@derobert I'm not sure exactly what I mean. See the "vague" part.
@derobert I'm having some trouble trying to get this recipe to work.
8
A: How to move / copy logical volume (lv) to another volume group (vg)?

Gabrielvgmerge lets you merge two VGs. You can also use pvmove to move data within a VG, and vgsplit if you want to go back to multiple VGs.

I want to copy, not move. But pvmove only moves.
I'm examining options. It seems that it might be possible to break a LVM mirror. Do you have any idea? It seems it wasn't possible once, but that might have changed.
Or are there any other alternatives?
?
I'm currently going with loop devices, but would consider other options.
Short story short - can I use lvconvert to create a mirror and then break it to make a new copy?
16:37
@FaheemMitha I think you can somehow...
@derobert So, am I missing anything?
   lvconvert --splitmirrors Number VG/RaidLV
   · Split images from raid1 RaidLV and use them to create a new LV.
   · Requires --name for the new LV, or the use of --trackchanges.
... from the lvconvert manual
I have no idea why that guide forces the old non-md mirror; manual seems to say that it works with both.
@derobert Yes, I see that. But I need the "other part" of the mirror to be created inside a specific physical extent (or device).
@derobert If you mean the link I posted, it might be out of date, or incorrect.
I had no idea that command existed till I read that page...
No, actually, I don't see what he is forcing.
@FaheemMitha I'm not sure if there is a way to override the default allocation policy when adding a mirror... trying to find it in the manpage. Of course, you could temporarily change the default policy.
@derobert There's an alloc option.
That man page isn't the chattiest thing around. I wish I could find a better guide.
An up-to-date one, naturally.
  Convert a linear LV to a two-way mirror LV, allocating new extents from specific PV ranges:
   lvconvert --mirrors 1 vg/lvol1 /dev/sda:0-15 /dev/sdb:0-15
This one might be relevant.
But I don't see documentation for the syntax. Sigh.
16:45
@FaheemMitha looks like you've found it...
@derobert I don't understand how it is supposed to work, though.
What's the syntax?
 /dev/sda:0-15 /dev/sdb:0-15
?
that's the PV name and the extent numbers, I think. If you look at the next example, you can give just the PV name and let it pick its own extent numbers.
Convert a mirror LV to a linear LV, freeing physical extents from a specific PV:
   lvconvert --type linear vg/lvol1 /dev/sda
This one? ^^
I don't really know what extent numbers are.
Oh. No, that's the other way around.
They're basically blocks for LVM. LVM stores its metadata in terms of extents. So a logical volume metadata includes a lists of PVs and the extents on each PV
look at the file in /etc/lvm/backup
@derobert Ah, blocks inside the VG?
16:49
                swap {
                        id = "anPe08-Tj1m-hBYK-Szaw-1oxi-JmGi-jn4qMD"
                        status = ["READ", "WRITE", "VISIBLE"]
                        flags = []
                        segment_count = 1

                        segment1 {
                                start_extent = 0
                                extent_count = 1430     # 5.58594 Gigabytes

                                type = "striped"
                                stripe_count = 1        # linear
that's an example from my machine. The LV "swap" starts at pv0, extent 0 and runs through extent 1430
@derobert Yes, I see.
Are these signposts relative to the containing/enclosing VG?
No, I think it's relative to the PV
@derobert Oh.
And if the VG contains multiple PVs?
Then it shouldn't be any different. The LV can contain multiple segments, each one specifies the next set of extents (and each segment can be on a different PV)
Or with striping, striped across multiple PVs.
@derobert Ah, ok.
So, getting back to my question. Do I need to specify a physical device when creating the mirror, or when splitting?
16:55
@FaheemMitha When creating. Otherwise, LVM will pick for you (and it should pick reasonably—i.e, a different PV than the source)
@derobert ok.
So, something like:
 lvconvert --type mirror --mirrors 1 vg/lvol1 /dev/sda
?
Yeah, though I'd use --type raid1 as I can't think of a reason to use the old non-md mirror code.
Or as the manpage puts it:
   In  most cases, the mirror type is deprecated and the raid1 type should
   be used.  They are both implementations of mirroring.
Ah, this has a detailed example:
@derobert Oh, ok. So
lvconvert --type raid1 --mirrors 1 vg/lvol1 /dev/sda
?
yep
presuming you want the copy on /dev/sda
Step 9 in that link has:
lvconvert -m 1 /dev/tecmint_vg/tecmint_lv /dev/sda1
17:00
also, is /dev/sda your PV? And not something like /dev/sda1?
which looks similar.
@derobert Oh, I'm just using whatever the documentation is doing.
I think my current devices are /dev/loop1, /dev/loop2. Test mode - loop devices. Does that look right?
And my actual devices are dev/md1, dev/md2, I think.
Or something like that.
Does what look right? The lvconvert --type raid1 --mirrors 1 «source-lv» «dest-pv» syntax looks right
pvs or pvdisplay will tell you the PV devices
@derobert ok
root@orwell:/home/faheem# pvs
  PV         VG        Fmt  Attr PSize   PFree
  /dev/loop1 testvg1   lvm2 a--   92.00m  80.00m
  /dev/loop2 testvg1   lvm2 a--   92.00m  80.00m
  /dev/md0   data      lvm2 a--  297.96g 137.96g
  /dev/md1   debian    lvm2 a--  465.76g  16.12g
  /dev/md127 newdebian lvm2 a--  465.40g 324.78g
Those first two entries are my test devices.
yeah, so for the test you'd use /dev/loop1 or /dev/loop2
@derobert Ok. Trying that now. Have you used any of this in the "wild"?
17:05
Though I don't think you've created your test LV yet, since both loop1 and loop2 show the same free space.
@FaheemMitha yes
Mainly via pvmove
@derobert I have. I don't know why it isn't showing up. Though they are basically empty.
I created one on each test vg.
@derobert What if you want to do a copy?
@FaheemMitha I haven't really needed to make copies that often.
@derobert Ok.
Except as part of moving data, like pvmove does.
Btw, suggestions for a better title?
The current one is awful.
@derobert Copying is conservative.
I mean, if something goes wrong, at least you (might) still have your source volume/filesystem.
When it comes to computers, I'm super-cautious.
Well, computer software.
17:09
@FaheemMitha Well, mostly when I've been doing that, it's been after replacing disks... and either it's being done by mdraid or alternatively I've built a new machine, then I copy from the old one using tar...
@derobert Yes, I thought I'd let LVM do the copying bit. Because it should be able to, IMO. I could have used dd or something. But I'm not a big fan of those raw write kinds of things.
Too little safety.
I think your LVM raid1 + split codepath is well-tested, so should be fine. It's pretty similar to how pvmove works, and that's definitely well-tested.
@derobert I'll try it. Thanks for the assistance. And maybe add an answer in a couple of places. This method isn't documented on SE, as far as I can see.
pvmoveis mentioned in a couple of places.
@derobert So, thoughts about test modes? Do you use loop devices or something else?
@FaheemMitha Loop devices should work. So would a VM. A VM would provide a more isolated test environment.
@derobert True.
Would a chroot work, or do you need something more isolated?
17:22
Isolated in both protecting the host from f— ups inside the VM, and also in keeping the test from being affected by weirdness on the host.
@FaheemMitha Depends on what you're testing!
@derobert Well, something like the stuff above.
Those commands are safe enough I'd test them on loop devices.
Being careful, of course, and making sure to use distinct names. A mistaken lvremove will still kill your real volumes
(though lvremove is undoable... as long as you catch it before your next lvcreate/lvextend)
@derobert Good to know. But for something like this I usually use cut and paste from a transcript.
Yes, a week or so ago I accidentally removed the wrong LV, and got to learn how to undo it. Turns out it's easy with vgcfgrestore
@derobert That must have been scary.
What about removing a volume group?
17:29
@FaheemMitha Not that bad, from how LVM works it clearly is recoverable, I was quite pleased with how easy it is though.
I was expecting to have to issue a scary lvcreate specifying exactly the same extents...
@derobert What command did you type?
Approximately, anyway.
@FaheemMitha I'm not sure exactly what vgremove does, but should be recoverable.
@FaheemMitha vgcfgrestore --file /etc/lvm/archive/Zia_00207-4082409.vg Zia ... pretty sure that was the command. Found it in history.
@derobert There's an archive?
17:32
LVM makes a backup of the metadata each time you change it
@derobert I'll keep that in mind...
Do you know if it ever gets rid of it?
Yeah. It keeps them for (at least the settings on my system are) at least 10 archives, and any under 30 days in age
... this consumes 324K on my machine
Well, more, actually since they're in /etc and I store /etc in git. But I have no idea to ask git how much space it takes, under any of the reasonably definitions of space used in a deduplicating store.
Search for the "backup" section on your /etc/lvm/lvm.conf
@derobert That seems quite comprehensive.
Since earlier settings usually quickly become irrelevant. Since the data gets overwritten, anyway.
 
1 hour later…
18:55
Does anyone know how to cut and paste with the mouse without introducing lots of extra crap. Some of it tabs, by the look of it.
Into Vim?
@Kusalananda From Konsole into Emacs. Sorry, I should have been more specific.
@FaheemMitha don't select crap?
@Gilles Not helpful.
@FaheemMitha if you want a clearer answer, ask a clearer question
19:02
@Gilles I'm cutting and pasting from a Konsole terminal. Very ordinary.
But I seem to be getting extra line endings or something.
19:45
BTW: I edited unix.stackexchange.com/questions/374282/… earlier today to make it clear—I think I succeeded, but it'd already been through the reopen queue so I don't think it gets another go. Needs two more votes to reopen.
@derobert One from me now.
It's open again. Thank you all.
@derobert the original version sure didn't make any sense
@Gilles no, OP was hopelessly confused to start
is he the same guy who was trolling (?) insisting to do all kinds of weird stuff by using cat a while back?
19:56
@Gilles Not sure. Doesn't show on his profile if so, but maybe it was deleted.
20:24
@derobert Well, that approach worked. Do you think it worth adding to the existing answers?
20:36
I've been getting the following by email since upgrade to stretch. I'm not sure to make of it. Should I report it as a bug, or did I somehow cause it? And is it a suitable question for U&L?
ubject: Anacron job 'cron.weekly' on orwell

/etc/cron.weekly/dhelp:
/usr/lib/ruby/2.3.0/rubygems/core_ext/kernel_require.rb:55:in `require': cannot load such file -- bdb (LoadError)
        from /usr/lib/ruby/2.3.0/rubygems/core_ext/kernel_require.rb:55:in `require'
        from /usr/lib/ruby/vendor_ruby/dhelp.rb:21:in `<top (required)>'
        from /usr/lib/ruby/2.3.0/rubygems/core_ext/kernel_require.rb:55:in `require'
        from /usr/lib/ruby/2.3.0/rubygems/core_ext/kernel_require.rb:55:in `require'

« first day (2456 days earlier)      last day (2796 days later) »